[ { "id": "DIV-0001", "case_id": "AIBD-0168", "title": "NCAR's DICast wind forecasts cut Xcel Energy's forecast error to 10.8%; Xcel estimates US$49 million saved", "summary": "saved through forecasting improvements, cumulative to 2015", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "US $49 million", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "USD", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "unresolved", "period": "Cumulative to 2015; the account places forecasting work from 2009, but the amount-specific start is unstated.", "unit": "cumulative savings across Xcel customers", "beneficiary": "Xcel Energy customers", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Appears to be gross avoided operating cost; implementation and running costs are unstated.", "comparator": "Operational forecast-error baseline; no comparable monetary baseline is stated.", "ai_attribution": "The estimate concerns forecasting improvements as a whole; the incremental AI share is not established.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://spectrum.ieee.org/taming-wind-power-with-better-forecasts", "title": "NCAR's DICast wind forecasts cut Xcel Energy's forecast error to 10.8%; Xcel estimates US$49 million saved", "quote": "Xcel estimates that improvements in forecasting have saved customers US $49 million so far.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Amount-specific start date", "Calculation method and baseline operating cost", "Implementation and ongoing costs", "AI component and its incremental contribution" ], "overlap_flags": [ "cumulative_not_annual", "whole_program_attribution", "ai_component_needs_review", "operator_estimate" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "Do not add the separate 2010 saving or prospective annual 2017 figure to the cumulative amount.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "electricity balancing and forecast-related operating costs", "claim_status": "retrospective_estimate", "currency": "USD", "currency_basis": "explicit", "unit_basis": "cumulative savings across Xcel customers", "time_basis": "Cumulative to 2015; the account places forecasting work from 2009, but the amount-specific start is unstated.", "time_basis_completeness": "incomplete_window", "monetary_comparator_assessment": "Operational forecast-error baseline; no comparable monetary baseline is stated.", "ai_and_value_scope": "The estimate concerns forecasting improvements as a whole; the incremental AI share is not established.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Appears to be gross avoided operating cost; implementation and running costs are unstated.", "net_benefit_established": false, "comparability_flags": [ "cumulative_not_annual", "whole_program_attribution", "ai_component_needs_review", "operator_estimate" ], "data_gaps": [ "Amount-specific start date", "Calculation method and baseline operating cost", "Implementation and ongoing costs", "AI component and its incremental contribution" ], "extraction_note": "Do not add the separate 2010 saving or prospective annual 2017 figure to the cumulative amount.", "source_review_required": true, "id": "AIBD-0168", "title": "NCAR's DICast wind forecasts cut Xcel Energy's forecast error to 10.8%; Xcel estimates US$49 million saved", "value": { "amount": "US $49 million", "what": "saved through forecasting improvements, cumulative to 2015", "who": "Xcel Energy customers", "estimate": true, "quote": "Xcel estimates that improvements in forecasting have saved customers US $49 million so far.", "url": "https://spectrum.ieee.org/taming-wind-power-with-better-forecasts" }, "sourceURL": "https://spectrum.ieee.org/taming-wind-power-with-better-forecasts", "date_observed_as_stored": "2015-10-22", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0002", "case_id": "AIBD-0186", "title": "CMS's Fraud Prevention System identifies or prevents $654.8 million in 2015, an $11.5-to-$1 return on investment", "summary": "inappropriate Medicare payments identified or prevented in calendar year 2015", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "$654.8 million", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "USD", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "avoided_loss", "amount_basis": "unresolved", "period": "Calendar year 2015", "unit": "program aggregate", "beneficiary": "Medicare program (CMS)", "payer": "Not established in the current source review.", "realization": "mixed", "accounting_boundary": "The benefit amount is distinct from the reported estimated $57.1 million cost; the headline amount is not a net benefit.", "comparator": "The existing comparator is the prior program year, not a no-AI counterfactual.", "ai_attribution": "FPS-supported investigations and actions; aggregate includes several accounting categories.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://web.archive.org/web/20160824134124id_/https://www.cms.gov/about-cms/components/cpi/downloads/fraud-prevention-system-return-on-investment-fourth-implementation-year-2015.pdf", "title": "CMS's Fraud Prevention System identifies or prevents $654.8 million in 2015, an $11.5-to-$1 return on investment", "quote": "The FPS helped identify or prevent $654.8 million in inappropriate payments during calendar year 2015 through actions taken due to the FPS or through investigations expedited, augmented, or corroborated by the FPS (Table 1).", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Separate realized recovery, prevented disbursement, projected future avoidance and referrals", "Full ROI cost boundary", "Contribution of AI versus investigation" ], "overlap_flags": [ "mixed_recovery_prevention_projection", "prior_year_not_counterfactual", "whole_program_attribution", "estimate_flag_too_coarse" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "Hold: the $654.8 million combines accounting states including prevention, projected savings, and amounts referred for recovery. Split components before curating a monetary benefit.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "Medicare improper payments identified or prevented", "claim_status": "mixed_realized_projected_and_referred", "currency": "USD", "currency_basis": "inferred_from_US_program", "unit_basis": "program aggregate", "time_basis": "Calendar year 2015", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "The existing comparator is the prior program year, not a no-AI counterfactual.", "ai_and_value_scope": "FPS-supported investigations and actions; aggregate includes several accounting categories.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "The benefit amount is distinct from the reported estimated $57.1 million cost; the headline amount is not a net benefit.", "net_benefit_established": false, "comparability_flags": [ "mixed_recovery_prevention_projection", "prior_year_not_counterfactual", "whole_program_attribution", "estimate_flag_too_coarse" ], "data_gaps": [ "Separate realized recovery, prevented disbursement, projected future avoidance and referrals", "Full ROI cost boundary", "Contribution of AI versus investigation" ], "extraction_note": "The existing estimate=false flag does not resolve the projected and referred components described in the account.", "source_review_required": true, "id": "AIBD-0186", "title": "CMS's Fraud Prevention System identifies or prevents $654.8 million in 2015, an $11.5-to-$1 return on investment", "value": { "amount": "$654.8 million", "what": "inappropriate Medicare payments identified or prevented in calendar year 2015", "who": "Medicare program (CMS)", "estimate": false, "quote": "The FPS helped identify or prevent $654.8 million in inappropriate payments during calendar year 2015 through actions taken due to the FPS or through investigations expedited, augmented, or corroborated by the FPS (Table 1).", "url": "https://web.archive.org/web/20160824134124id_/https://www.cms.gov/about-cms/components/cpi/downloads/fraud-prevention-system-return-on-investment-fourth-implementation-year-2015.pdf" }, "sourceURL": "https://web.archive.org/web/20160824134124id_/https://www.cms.gov/about-cms/components/cpi/downloads/fraud-prevention-system-return-on-investment-fourth-implementation-year-2015.pdf", "date_observed_as_stored": "2015-12-31", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0003", "case_id": "AIBD-0196", "title": "Global Fishing Watch data lets Kiribati fine a vessel US$1 million for fishing in the Phoenix Islands Protected Area", "summary": "fine secured by Kiribati against the owner of Marshalls 203 for fishing illegally in the Phoenix Islands Protected Area, plus an additional $1 million paid as a grant under a goodwill arrangement", "monetary_role": "recipient_transfer", "amount": { "source_text": "$1 million", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "USD", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "recipient_transfer", "amount_basis": "unresolved", "period": "Single case reported by 2016; exact payment date absent.", "unit": "one enforcement settlement with two differently described payments", "beneficiary": "Government of Kiribati", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross government receipts or commitments; enforcement costs and losses to the payer are not included.", "comparator": "No monetary comparator.", "ai_attribution": "Tracking supported enforcement; classifier contribution and payment realization remain unclear.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://grist.org/science/this-new-digital-tool-could-help-fight-overfishing-around-the-world/", "title": "Global Fishing Watch data lets Kiribati fine a vessel US$1 million for fishing in the Phoenix Islands Protected Area", "quote": "This helped the country secure a $1 million fine against the Central Pacific Fishing Company, the owner of Marshalls 203. The company also made a special “goodwill arrangement” with the government, agreeing to pay an additional $1 million in the form of a grant.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Proof and dates of payment", "Separate fine from grant", "Reconcile later reporting that describes a $2 million fine", "Counterfactual enforcement outcome" ], "overlap_flags": [ "transfer_not_total_social_value", "multiple_payment_components", "fine_grant_conflict", "ai_increment_unresolved" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "The amount field is $1 million, while what/quote also describe a separate $1 million grant. Do not silently treat the sum as one fine.", "reviewed_on": null, "legacy_audit": { "monetary_role": "recipient_transfer", "money_type": "other", "money_subtype": "enforcement fine and separate goodwill grant", "claim_status": "fine_secured_and_grant_agreed", "currency": "USD", "currency_basis": "currency_inferred_from_context", "unit_basis": "one enforcement settlement with two differently described payments", "time_basis": "Single case reported by 2016; exact payment date absent.", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "No monetary comparator.", "ai_and_value_scope": "Tracking supported enforcement; classifier contribution and payment realization remain unclear.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross government receipts or commitments; enforcement costs and losses to the payer are not included.", "net_benefit_established": false, "comparability_flags": [ "transfer_not_total_social_value", "multiple_payment_components", "fine_grant_conflict", "ai_increment_unresolved" ], "data_gaps": [ "Proof and dates of payment", "Separate fine from grant", "Reconcile later reporting that describes a $2 million fine", "Counterfactual enforcement outcome" ], "extraction_note": "The amount field is $1 million, while what/quote also describe a separate $1 million grant. Do not silently treat the sum as one fine.", "source_review_required": true, "id": "AIBD-0196", "title": "Global Fishing Watch data lets Kiribati fine a vessel US$1 million for fishing in the Phoenix Islands Protected Area", "value": { "amount": "$1 million", "what": "fine secured by Kiribati against the owner of Marshalls 203 for fishing illegally in the Phoenix Islands Protected Area, plus an additional $1 million paid as a grant under a goodwill arrangement", "who": "Government of Kiribati", "estimate": false, "quote": "This helped the country secure a $1 million fine against the Central Pacific Fishing Company, the owner of Marshalls 203. The company also made a special “goodwill arrangement” with the government, agreeing to pay an additional $1 million in the form of a grant.", "url": "https://grist.org/science/this-new-digital-tool-could-help-fight-overfishing-around-the-world/" }, "sourceURL": "https://grist.org/science/this-new-digital-tool-could-help-fight-overfishing-around-the-world/", "date_observed_as_stored": "2016-09-15", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0004", "case_id": "AIBD-0331", "title": "TCU's Alice robot flags DNIT notices for BR-153 maintenance on stretches already under concession; TCU Acórdão 1116/2017 cancels the tender", "summary": "value of the two DNIT tender notices for BR-153 maintenance that a TCU acórdão blocked", "monetary_role": "reference_or_input_amount", "amount": { "source_text": "R$ 39 milhões", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "BRL", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "reference_amount", "amount_basis": "reference", "period": "Tenders reported in September 2017; precise occurrence dates absent.", "unit": "two tender notices", "beneficiary": "DNIT's Goiás unit, which had published the tenders for stretches whose maintenance falls to the concessionaire", "payer": "Not established in the current source review.", "realization": "context", "accounting_boundary": "A contract or budget value is not a gross or net benefit.", "comparator": "No expenditure reduction comparator.", "ai_attribution": "AI flagged tenders, but the amount describes their value rather than a demonstrated monetary gain.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://web.archive.org/web/20240224093756/https://inforchannel.com.br/2017/09/27/tcu-cria-robo-para-varredura-diaria-de-irregularidades-em-editais/", "title": "TCU's Alice robot flags DNIT notices for BR-153 maintenance on stretches already under concession; TCU Acórdão 1116/2017 cancels the tender", "quote": "Em Goiás, o Tribunal de Contas, via Alice, descobriu que o DNIT estdual publicou dois editais, de R$ 39 milhões, para manutenção da BR 153, mas em trechos que estão sob concessão – portanto com manutenção a cargo da concessionária.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Reconcile R$39 million across two notices with R$20 million for one tender", "Original court/audit decision", "What payments actually would have occurred and were avoided" ], "overlap_flags": [ "contract_value_not_saving", "conflicting_scope_and_amount", "ai_increment_unresolved" ], "curation_status": "context_only", "source_review_status": "context_only", "case_link_status": "unresolved", "review_note": "Keep as a contextual procurement amount until a distinct saving or avoided payment is supported.", "reviewed_on": null, "legacy_audit": { "monetary_role": "reference_or_input_amount", "money_type": "other", "money_subtype": "face value of blocked procurement notices", "claim_status": "procurement_value_at_issue", "currency": "BRL", "currency_basis": "explicit", "unit_basis": "two tender notices", "time_basis": "Tenders reported in September 2017; precise occurrence dates absent.", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "No expenditure reduction comparator.", "ai_and_value_scope": "AI flagged tenders, but the amount describes their value rather than a demonstrated monetary gain.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "A contract or budget value is not a gross or net benefit.", "net_benefit_established": false, "comparability_flags": [ "contract_value_not_saving", "conflicting_scope_and_amount", "ai_increment_unresolved" ], "data_gaps": [ "Reconcile R$39 million across two notices with R$20 million for one tender", "Original court/audit decision", "What payments actually would have occurred and were avoided" ], "extraction_note": "Keep as a contextual procurement amount until a distinct saving or avoided payment is supported.", "source_review_required": true, "id": "AIBD-0331", "title": "TCU's Alice robot flags DNIT notices for BR-153 maintenance on stretches already under concession; TCU Acórdão 1116/2017 cancels the tender", "value": { "amount": "R$ 39 milhões", "what": "value of the two DNIT tender notices for BR-153 maintenance that a TCU acórdão blocked", "who": "DNIT's Goiás unit, which had published the tenders for stretches whose maintenance falls to the concessionaire", "estimate": false, "quote": "Em Goiás, o Tribunal de Contas, via Alice, descobriu que o DNIT estdual publicou dois editais, de R$ 39 milhões, para manutenção da BR 153, mas em trechos que estão sob concessão – portanto com manutenção a cargo da concessionária.", "url": "https://web.archive.org/web/20240224093756/https://inforchannel.com.br/2017/09/27/tcu-cria-robo-para-varredura-diaria-de-irregularidades-em-editais/" }, "sourceURL": "https://web.archive.org/web/20240224093756/https://inforchannel.com.br/2017/09/27/tcu-cria-robo-para-varredura-diaria-de-irregularidades-em-editais/", "date_observed_as_stored": "2017-09-27", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0005", "case_id": "AIBD-0105", "title": "Monetized climate value of GLAD alerts", "summary": "Researchers value avoided deforestation associated with African GLAD alert subscriptions at US$149–696 million using the social cost of carbon.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "US$149–696 million", "value": null, "lower": "149000000", "upper": "696000000", "qualifier": "range", "currency": "USD", "currency_basis": "explicit", "price_basis": "Social-cost-of-carbon valuation; valuation and price year are not specified in the checked abstract." }, "additional_amounts": [], "financial_type": "monetized_social_benefit", "amount_basis": "valuation", "period": "Comparison with pre-2016 deforestation; the checked abstract does not specify the valuation horizon.", "unit": "Modeled value of avoided deforestation in the African study sample.", "beneficiary": "People benefiting from avoided climate damage; no direct cash recipient.", "payer": "Not applicable to a social-cost valuation.", "realization": "modeled_valuation", "accounting_boundary": "Modeled climate value, not cash income, market revenue, or a fully costed net-benefit estimate.", "comparator": "Deforestation associated with subscriptions versus the study counterfactual. Alert availability alone had no significant effect.", "ai_attribution": "GLAD’s machine-learning forest-change alerts are inputs to the intervention. The evaluation concerns subscriptions and subsequent responses, not model performance in isolation.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://ageconsearch.umn.edu/record/304666/files/18339%20manuscript%20ABSTRACT%20ONLY.pdf", "title": "Near Real-Time Forest Monitoring Can Decrease Deforestation", "quote": "we estimate a value of the alert system between USD 149 million and 696 million.", "date": null, "primary": true, "written_by": "unknown", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "Author abstract, single-page PDF" } ], "limitations": [ "Only the author abstract was checked for this monetary range; the full valuation method remains unreviewed.", "The reported deforestation effect was specific to Africa, with no detected effect on other continents." ], "overlap_flags": [ "The interval represents alternative valuation assumptions, not two additive benefits." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "other", "money_subtype": "monetized environmental and social benefit", "claim_status": "retrospective_modeled_valuation", "currency": "USD", "currency_basis": "explicit", "unit_basis": "avoided deforestation in the African study region", "time_basis": "Study includes a 2011-2016 baseline; the valuation window is not explicit in value.", "time_basis_completeness": "incomplete_window", "monetary_comparator_assessment": "The account describes a counterfactual for deforestation; no cash-flow comparator.", "ai_and_value_scope": "Valuation concerns the alert system with human uptake and response, not AI in isolation.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Social-cost-of-carbon valuation; net costs and broader welfare effects are not supplied.", "net_benefit_established": false, "comparability_flags": [ "nonmarket_valuation", "range_not_point_estimate", "valuation_assumption_sensitive", "whole_program_attribution" ], "data_gaps": [ "Valuation period", "Carbon quantity and social cost of carbon assumptions", "Price year and discount rate", "Program costs and uncertainty method" ], "extraction_note": "This is a monetary valuation of an environmental outcome, not cash saved or revenue received.", "source_review_required": true, "id": "AIBD-0105", "title": "GLAD deforestation alert subscriptions reduce deforestation probability in Africa by 18%", "value": { "amount": "between US$149 million and US$696 million", "what": "value of avoided deforestation in Africa, priced at the social cost of carbon", "who": "society, through avoided carbon emissions", "estimate": true, "quote": "Calculated using the social cost of carbon for avoided deforestation in Africa, we estimate the alert system’s value to be between US$149 million and US$696 million.", "url": "https://www.nature.com/articles/s41558-020-00956-w" }, "sourceURL": "https://www.nature.com/articles/s41558-020-00956-w", "date_observed_as_stored": "2021-01-04", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0006", "case_id": "AIBD-0193", "title": "Avoided East African locust losses", "summary": "FAO estimates that the locust response avoided $1.56 billion in cereal and milk losses.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "US$1.56 billion", "value": "1560000000", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "explicit", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "avoided_loss", "amount_basis": "avoided_expenditure", "period": "East African response reported May 6, 2021; the amount-specific start date is unstated.", "unit": "Cereal and milk losses avoided across the regional response.", "beneficiary": "Farmers, livestock keepers, and food consumers in affected East African countries.", "payer": "Not applicable to avoided production losses.", "realization": "retrospective_estimate", "accounting_boundary": "Estimated gross production losses avoided by the response; response costs are not deducted.", "comparator": "Estimated losses without the control response; calculation details are not given in this account.", "ai_attribution": "PlantVillage AI helped clean monitoring data within a larger response involving field teams, aircraft, forecasts, and pest control. The AI-specific increment is unknown.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://www.fao.org/newsroom/story/A-high-tech-response-is-helping-countries-win-the-battle-against-Desert-Locusts/en", "title": "A high-tech response is helping countries win the battle against Desert Locusts", "quote": "avoiding USD 1.56 billion in cereal and milk losses.", "date": "2021-05-06", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": null } ], "limitations": [ "FAO reports a whole-response counterfactual estimate, not cash received or an independently isolated AI effect." ], "overlap_flags": [ "The cereal and milk quantities describe the same avoided-loss estimate and are not additional monetary gains." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "agricultural production losses", "claim_status": "retrospective_counterfactual_estimate", "currency": "USD", "currency_basis": "explicit", "unit_basis": "entire East African locust-control campaign", "time_basis": "Campaign beginning in 2020, reported in May 2021; precise valuation cutoff needs checking.", "time_basis_completeness": "incomplete_window", "monetary_comparator_assessment": "Assumed no-control-operations counterfactual; no quantified monetary baseline in the structured comparator.", "ai_and_value_scope": "The amount belongs to the whole control operation, including aircraft and field action, to which AI-supported information contributed.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross avoided production loss; campaign costs are not deducted.", "net_benefit_established": false, "comparability_flags": [ "whole_program_attribution", "counterfactual_estimate", "agricultural_valuation", "not_AI_only" ], "data_gaps": [ "Valuation method and price year", "Precise campaign period and geographic coverage", "AI contribution", "Cost of control operations" ], "extraction_note": "The amount must not be described as the app or AI alone generating USD 1.56 billion.", "source_review_required": true, "id": "AIBD-0193", "title": "eLocust3m and AI data cleaning feed FAO's East Africa locust control, which may have averted USD 1.56 billion in losses", "value": { "amount": "USD 1.56 billion", "what": "cereal and milk losses avoided by the locust control operations in East Africa that the app and AI data cleaning fed", "who": "East African farmers and herders (the food security of 36.6 million people)", "estimate": true, "quote": "In East Africa locust control operations have prevented the loss of 4 million tonnes of cereal and 800 million litres of milk production, while protecting the food security of 36.6 million people and avoiding USD 1.56 billion in cereal and milk losses.", "url": "https://www.fao.org/newsroom/story/A-high-tech-response-is-helping-countries-win-the-battle-against-Desert-Locusts/en" }, "sourceURL": "https://www.fao.org/newsroom/story/A-high-tech-response-is-helping-countries-win-the-battle-against-Desert-Locusts/en", "date_observed_as_stored": "2021-05-06", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0007", "case_id": "AIBD-0337", "title": "myTax prompts and estimated tax revenue", "summary": "ATO estimates that adjustments following myTax prompts affected 2020 tax revenue by about A$37 million.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "About A$37 million", "value": "37000000", "lower": null, "upper": null, "qualifier": "approximately", "currency": "AUD", "currency_basis": "context", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "revenue", "amount_basis": "incremental_change", "period": "Australian tax time 2020.", "unit": "Aggregate adjustments following myTax prompts.", "beneficiary": "Australian public revenue.", "payer": "Taxpayers adjusting their returns.", "realization": "retrospective_estimate", "accounting_boundary": "Estimated tax-revenue impact, not confirmed cash collection, net program return, or aggregate social benefit.", "comparator": "Amounts entered before versus after prompts; an untreated causal comparison is not supplied.", "ai_attribution": "ATO describes nearest-neighbor comparisons and other checks that prompted review. Their separate contributions are not quantified.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://www.ato.gov.au/api/public/content/0-55fbf5c1-cd9e-49b9-8c8a-42b607e83a36", "title": "ATO: How we use data and analytics", "quote": "These prompts caused taxpayers to make adjustments estimated to have a revenue impact of around $37 million.", "date": null, "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "indexed_excerpt", "locator": "Helping you make better choices" } ], "limitations": [ "The relevant official excerpt was available through search indexing; direct retrieval returned an access error.", "The source states an estimated revenue impact rather than collected taxes." ], "overlap_flags": [ "The separate $42 million of automated franking-credit refunds is not part of this estimate." ], "curation_status": "curated", "source_review_status": "primary_excerpt_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "revenue", "money_subtype": "tax assessment or revenue impact", "claim_status": "retrospective_estimate", "currency": "AUD", "currency_basis": "inferred_from_Australian_program", "unit_basis": "taxpayer adjustments during one tax season", "time_basis": "Tax time 2020", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "No separate control or monetary baseline is stated.", "ai_and_value_scope": "Attributed by ATO to adjustments following prompts; collection and incremental causality are not established.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Estimated gross revenue impact; not demonstrated net collected cash.", "net_benefit_established": false, "comparability_flags": [ "assessed_not_collected", "currency_inferred", "operator_estimate", "no_control" ], "data_gaps": [ "Confirmed currency", "Tax assessments versus actual receipts", "Counterfactual changes without prompts", "Program costs" ], "extraction_note": "Exclude unrelated refund totals in the account from this claim.", "source_review_required": true, "id": "AIBD-0337", "title": "ATO's nearest-neighbor myTax prompts reached nearly 340,000 taxpayers at tax time 2020; resulting adjustments had an estimated $37 million revenue impact", "value": { "amount": "$37 million", "what": "revenue impact of the adjustments taxpayers made after myTax prompts at tax time 2020 (ATO's own estimate)", "who": "the Australian Taxation Office (Australian Government revenue)", "estimate": true, "quote": "During tax time in 2020, nearly 340,000 taxpayers (around 7.5% of myTax users), received a pop-up message through myTax suggesting they review a specific label. These prompts caused taxpayers to make adjustments estimated to have a revenue impact of around $37 million.", "url": "https://web.archive.org/web/20210714025448/https://www.ato.gov.au/About-ATO/Commitments-and-reporting/Information-and-privacy/How-we-use-data-and-analytics/" }, "sourceURL": "https://web.archive.org/web/20210714025448/https://www.ato.gov.au/About-ATO/Commitments-and-reporting/Information-and-privacy/How-we-use-data-and-analytics/", "date_observed_as_stored": "2021-07-14", "existing_estimate_flag": true, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0008", "case_id": "AIBD-0082", "title": "CGU, supported by ALICE, cuts R$2.38 million from a 2021 Hospital Federal de Ipanema glove tender and R$5.95 million from an undated INCA glove tender", "summary": "overpriced hospital purchases avoided across five preventive audits", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "R$ 8,33 milhões", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "BRL", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "unresolved", "period": "Reported October 2023; at least one component is dated 2021; full interval is unclear.", "unit": "five preventive audits of Rio federal hospitals and institutes", "beneficiary": "the federal government (hospitals and institutes in Rio de Janeiro)", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Appears to be gross cost avoidance; paid-price outcomes and audit costs are not established.", "comparator": "Original and revised estimates appear for individual components, not a fully reconciled aggregate.", "ai_attribution": "Human audit and procurement revisions followed ALICE alerts.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://web.archive.org/web/20260412063523id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2023/10/cgu-atua-em-editais-de-licitacao-em-hospitais-federais-no-rio-de-janeiro-e-evita-desperdicio-de-cerca-de-r-8-33-mi", "title": "CGU, supported by ALICE, cuts R$2.38 million from a 2021 Hospital Federal de Ipanema glove tender and R$5.95 million from an undated INCA glove tender", "quote": "A Controladoria-Geral da União (CGU) realizou, preventivamente, cinco auditorias em editais de contratações de hospitais federais e institutos sediados no Rio de Janeiro/RJ. A ação evitou que, pelo menos, R$ 8,33 milhões fossem desperdiçados em aquisições de insumos e serviços hospitalares com sobrepreço.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Reconcile all five audit components", "Full event interval", "Final contract/payment amounts", "Audit and system costs" ], "overlap_flags": [ "budget_revision_not_cash_release", "multi_case_aggregate", "component_baselines_only", "possible_overlap_with_program_aggregates" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "Preserve the lower-bound language of at least R$8.33 million. Do not substitute the full original-versus-auction price gap for the cited benefit.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "procurement cost avoidance", "claim_status": "reported_preventive_audit_saving", "currency": "BRL", "currency_basis": "explicit", "unit_basis": "five preventive audits of Rio federal hospitals and institutes", "time_basis": "Reported October 2023; at least one component is dated 2021; full interval is unclear.", "time_basis_completeness": "incomplete_window", "monetary_comparator_assessment": "Original and revised estimates appear for individual components, not a fully reconciled aggregate.", "ai_and_value_scope": "Human audit and procurement revisions followed ALICE alerts.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Appears to be gross cost avoidance; paid-price outcomes and audit costs are not established.", "net_benefit_established": false, "comparability_flags": [ "budget_revision_not_cash_release", "multi_case_aggregate", "component_baselines_only", "possible_overlap_with_program_aggregates" ], "data_gaps": [ "Reconcile all five audit components", "Full event interval", "Final contract/payment amounts", "Audit and system costs" ], "extraction_note": "Preserve the lower-bound language of at least R$8.33 million. Do not substitute the full original-versus-auction price gap for the cited benefit.", "source_review_required": true, "id": "AIBD-0082", "title": "CGU, supported by ALICE, cuts R$2.38 million from a 2021 Hospital Federal de Ipanema glove tender and R$5.95 million from an undated INCA glove tender", "value": { "amount": "R$ 8,33 milhões", "what": "overpriced hospital purchases avoided across five preventive audits", "who": "the federal government (hospitals and institutes in Rio de Janeiro)", "estimate": false, "quote": "A Controladoria-Geral da União (CGU) realizou, preventivamente, cinco auditorias em editais de contratações de hospitais federais e institutos sediados no Rio de Janeiro/RJ. A ação evitou que, pelo menos, R$ 8,33 milhões fossem desperdiçados em aquisições de insumos e serviços hospitalares com sobrepreço.", "url": "https://web.archive.org/web/20260412063523id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2023/10/cgu-atua-em-editais-de-licitacao-em-hospitais-federais-no-rio-de-janeiro-e-evita-desperdicio-de-cerca-de-r-8-33-mi" }, "sourceURL": "https://web.archive.org/web/20260412063523id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2023/10/cgu-atua-em-editais-de-licitacao-em-hospitais-federais-no-rio-de-janeiro-e-evita-desperdicio-de-cerca-de-r-8-33-mi", "date_observed_as_stored": "2021-07-21", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0009", "case_id": "AIBD-0317", "title": "Vyntra's (formerly NetGuardians) AI risk models may have blocked fraudulent payments at unnamed banks, including a USD 500,000 ATM scheme", "summary": "fraudulent ATM withdrawal detected and stopped, with arrests", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "USD 500,000", "value": "500000", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [ { "id": "kes-equivalent", "amount": { "source_text": "KSh 58 million", "value": "58000000", "lower": null, "upper": null, "qualifier": "exact", "currency": "KES", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "relationship": "source_equivalent", "equivalent_to": "primary", "equivalence_note": "The source gives both currencies for the same amount. It does not specify an exchange-rate date or method.", "period": "No event date established", "unit": "one attempted withdrawal at an unnamed bank", "realization": "unverified", "accounting_boundary": "Nominal attempted amount, before intervention costs; expected loss without intervention is unknown.", "sources": [ { "url": "https://vyntra.com/fraud-cases/", "title": "Fraud Cases - Vyntra", "quote": "USD 500,000 (KSh 58 million)", "date": null, "primary": true, "written_by": "makers", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "Kenya - Internal and channels fraud" } ], "source_review_status": "primary_source_checked", "review_note": "The currency expression is explicit in the source. Its equivalence does not establish a separate gain or resolve the claim’s remaining evidence gaps." } ], "financial_type": "avoided_loss", "amount_basis": "unresolved", "period": "No event date established", "unit": "one attempted withdrawal at an unnamed bank", "beneficiary": "the second bank, an unnamed client bank (the page labels the case 'Kenya')", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Nominal attempted amount, before intervention costs; expected loss without intervention is unknown.", "comparator": "No independent counterfactual or expected-loss estimate.", "ai_attribution": "Vendor attributes the intervention to its models; AI and human contributions are not separated.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://vyntra.com/fraud-cases/", "title": "Fraud Cases - Vyntra", "quote": "The fraud was detected in real time and the fraudsters were arrested while attempting to withdraw USD 500,000 (KSh 58 million) from the second bank’s ATMs. Vyntra's AI risk models prevented the fraud due to the unusual timing and frequency of ATM withdrawals.", "date": null, "primary": true, "written_by": "makers", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "Kenya - Internal and channels fraud" } ], "limitations": [ "Bank identity and confirmation", "Event date", "Expected loss or later recovery without the system", "AI mechanism and intervention cost" ], "overlap_flags": [ "vendor_claim", "dual_currency_same_amount", "undated_case", "attempted_amount_not_expected_loss" ], "curation_status": "awaiting_review", "source_review_status": "source_checked_link_unresolved", "case_link_status": "unresolved", "review_note": "USD 500,000 and KSh 58 million describe the same attempted withdrawal. The bank and event date remain unnamed, and the amount does not establish the loss that would otherwise have occurred.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "attempted fraudulent withdrawal stopped", "claim_status": "reported_prevention_of_nominal_attempt", "currency": "USD; KES equivalent", "currency_basis": "explicit", "unit_basis": "one attempted withdrawal at an unnamed bank", "time_basis": "No event date established", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "No independent counterfactual or expected-loss estimate.", "ai_and_value_scope": "Vendor attributes the intervention to its models; AI and human contributions are not separated.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Nominal attempted amount, before intervention costs; expected loss without intervention is unknown.", "net_benefit_established": false, "comparability_flags": [ "vendor_claim", "dual_currency_same_amount", "undated_case", "attempted_amount_not_expected_loss" ], "data_gaps": [ "Bank identity and confirmation", "Event date", "Expected loss or later recovery without the system", "AI mechanism and intervention cost" ], "extraction_note": "USD 500,000 and KSh 58 million describe the same amount and must not be added.", "source_review_required": true, "id": "AIBD-0317", "title": "Vyntra's (formerly NetGuardians) AI risk models may have blocked fraudulent payments at unnamed banks, including a USD 500,000 ATM scheme", "value": { "amount": "USD 500,000 (KSh 58 million)", "what": "fraudulent ATM withdrawal detected and stopped, with arrests", "who": "the second bank, an unnamed client bank (the page labels the case 'Kenya')", "estimate": false, "quote": "The fraud was detected in real time and the fraudsters were arrested while attempting to withdraw USD 500,000 (KSh 58 million) from the second bank’s ATMs. Vyntra's AI risk models prevented the fraud due to the unusual timing and frequency of ATM withdrawals.", "url": "https://www.vyntra.com/fraud-cases/" }, "sourceURL": "https://www.vyntra.com/fraud-cases/", "date_observed_as_stored": "2022-04-04", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0010", "case_id": "AIBD-0031", "title": "Lower household medical spending after flood alerts", "summary": "A randomized study estimates that community dissemination of AI flood alerts reduced household illness-treatment spending by roughly INR6,500.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "About INR6,500 per household", "value": "6500", "lower": null, "upper": null, "qualifier": "approximately", "currency": "INR", "currency_basis": "explicit", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "incremental_change", "period": "2022–2023 flood-season study; the prose does not isolate the monetary recall window.", "unit": "Estimated difference per household in illness-treatment spending.", "beneficiary": "Households in treatment communities in Bihar, India.", "payer": "Households paying for illness treatment.", "realization": "retrospective_estimate", "accounting_boundary": "Estimated household medical-expenditure reduction, not a full net-welfare or program-profit estimate.", "comparator": "Randomized community dissemination versus control communities, both with access to the AI forecasting system.", "ai_attribution": "The randomized intervention is dissemination of AI forecasts, not AI versus no AI. Its effect cannot be relabeled as the forecasting model’s isolated contribution.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://bpb-us-w2.wpmucdn.com/campuspress.yale.edu/dist/7/2986/files/2024/04/Google_Floods-bcdf49042e23059a.pdf", "title": "Forecasting Fate: Experimental Evaluation of a Flood Early Warning System", "quote": "households in treatment areas incurred roughly INR 6,500 less on illness treatment", "date": "2024-04-03", "primary": true, "written_by": "independent", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "PDF page 6; medical-expenditure result and Table 4" } ], "limitations": [ "The manuscript reports preliminary estimates and no evidence of broader improvements in economic well-being.", "The reported amount is a study estimate, not a guaranteed saving for each household." ], "overlap_flags": [ "The paper’s benefit-cost ratio uses these medical savings; do not count it as another benefit." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "household medical expenditure", "claim_status": "study_estimate_of_observed_spending_difference", "currency": "INR", "currency_basis": "explicit", "unit_basis": "per household over a flood season", "time_basis": "2022 and 2023 monsoon study; the amount-specific season requires confirmation.", "time_basis_completeness": "partial_window", "monetary_comparator_assessment": "Randomized comparison for community dissemination; control households already had access to AI phone alerts.", "ai_and_value_scope": "Incremental effect of community dissemination added to forecast access, not AI forecasts versus no AI.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Household expenditure saving before program costs; a separate benefit-cost ratio does not make the headline amount net.", "net_benefit_established": false, "comparability_flags": [ "per_household", "study_estimate_despite_false_flag", "comparator_already_has_AI", "season_specific", "increment_is_delivery_layer" ], "data_gaps": [ "Which season and sample support INR 6,500", "Confidence interval and attrition", "How illness spending was measured", "Program cost and ratio assumptions" ], "extraction_note": "A measured study estimate still has sampling uncertainty; estimate=false should not imply an exact cash-accounting total.", "source_review_required": true, "id": "AIBD-0031", "title": "Randomized evaluation of Google AI flood alerts with community dissemination in Bihar, India (2022–2023 monsoons)", "value": { "amount": "INR 6,500", "what": "less spent on illness treatment per household over the flood season", "who": "treatment households in the Bihar study panchayats", "estimate": false, "quote": "As a result, households in treatment areas incurred roughly INR 6,500 less on illness treatment (a 32.87% decrease). In fact, based on the estimated reduction in medical expenditures alone, we estimate that for every $1 spent on the program, the intervention would generate a minimum benefit of $30 during a severe flooding season.", "url": "https://bpb-us-w2.wpmucdn.com/campuspress.yale.edu/dist/7/2986/files/2024/04/Google_Floods-bcdf49042e23059a.pdf" }, "sourceURL": "https://bpb-us-w2.wpmucdn.com/campuspress.yale.edu/dist/7/2986/files/2024/04/Google_Floods-bcdf49042e23059a.pdf", "date_observed_as_stored": "2022-06-01", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0011", "case_id": "AIBD-0332", "title": "TCE-MG's Alice audit of 2,297 August 2022 tenders detects 60 irregularities worth R$113.7 million; 53 alerts answered and notices worth R$91.7 million corrected", "summary": "value of the tender notices corrected after Alice's alerts in the audit of August 2022 tenders", "monetary_role": "reference_or_input_amount", "amount": { "source_text": "R$ 91.723.011,31", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "BRL", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "reference_amount", "amount_basis": "reference", "period": "August 2022 tender cohort; audit described in December 2022 and reported June 2023.", "unit": "corrected notices in one audit cohort", "beneficiary": "the audited state and municipal bodies in Minas Gerais, which republished corrected notices; the page's title calls the result a saving without saying who saved", "payer": "Not established in the current source review.", "realization": "context", "accounting_boundary": "Value of corrected notices, not a gross or net saving.", "comparator": "No actual expenditure reduction or monetary counterfactual.", "ai_attribution": "ALICE alerts supported corrections; no benefit amount is isolated.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://atricon.org.br/blog/tce-mg-apresenta-economia-feita-com-ajuda-de-robo-alice/", "title": "TCE-MG's Alice audit of 2,297 August 2022 tenders detects 60 irregularities worth R$113.7 million; 53 alerts answered and notices worth R$91.7 million corrected", "quote": "Dos 60 alertas encaminhados aos jurisdicionados, 53 apresentaram respostas que permitiram a verificação do controle, alcançando a correção das irregularidades na republicação dos editais ou adoção de providências em futuros procedimentos licitatórios. Com isso, a auditoria conseguiu a correção dos editais equivalente a R$ 91.723.011,31.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Amounts changed, canceled, saved or recovered", "Final procurement outcomes", "Separate corrected-contract value from benefit value" ], "overlap_flags": [ "contract_value_not_saving", "headline_accounting_mismatch", "audit_cohort" ], "curation_status": "context_only", "source_review_status": "context_only", "case_link_status": "unresolved", "review_note": "The existing amount explicitly measures the tenders corrected; the source headline calling it a saving is insufficient to change its meaning.", "reviewed_on": null, "legacy_audit": { "monetary_role": "reference_or_input_amount", "money_type": "other", "money_subtype": "face value of corrected tenders", "claim_status": "procurement_value_corrected", "currency": "BRL", "currency_basis": "explicit", "unit_basis": "corrected notices in one audit cohort", "time_basis": "August 2022 tender cohort; audit described in December 2022 and reported June 2023.", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "No actual expenditure reduction or monetary counterfactual.", "ai_and_value_scope": "ALICE alerts supported corrections; no benefit amount is isolated.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Value of corrected notices, not a gross or net saving.", "net_benefit_established": false, "comparability_flags": [ "contract_value_not_saving", "headline_accounting_mismatch", "audit_cohort" ], "data_gaps": [ "Amounts changed, canceled, saved or recovered", "Final procurement outcomes", "Separate corrected-contract value from benefit value" ], "extraction_note": "The existing amount explicitly measures the tenders corrected; the source headline calling it a saving is insufficient to change its meaning.", "source_review_required": true, "id": "AIBD-0332", "title": "TCE-MG's Alice audit of 2,297 August 2022 tenders detects 60 irregularities worth R$113.7 million; 53 alerts answered and notices worth R$91.7 million corrected", "value": { "amount": "R$ 91.723.011,31", "what": "value of the tender notices corrected after Alice's alerts in the audit of August 2022 tenders", "who": "the audited state and municipal bodies in Minas Gerais, which republished corrected notices; the page's title calls the result a saving without saying who saved", "estimate": false, "quote": "Dos 60 alertas encaminhados aos jurisdicionados, 53 apresentaram respostas que permitiram a verificação do controle, alcançando a correção das irregularidades na republicação dos editais ou adoção de providências em futuros procedimentos licitatórios. Com isso, a auditoria conseguiu a correção dos editais equivalente a R$ 91.723.011,31.", "url": "https://atricon.org.br/blog/tce-mg-apresenta-economia-feita-com-ajuda-de-robo-alice/" }, "sourceURL": "https://atricon.org.br/blog/tce-mg-apresenta-economia-feita-com-ajuda-de-robo-alice/", "date_observed_as_stored": "2022-12-01", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0012", "case_id": "AIBD-0116", "title": "York student has a £60 parking fine withdrawn after sending an appeal letter drafted by ChatGPT", "summary": "parking fine withdrawn by York City Council after the ChatGPT-drafted appeal", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "£60", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "GBP", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "unresolved", "period": "Case reported March 2023; exact withdrawal date absent.", "unit": "one parking fine", "beneficiary": "Millie Houlton", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross personal payment avoided; transfer effect for the council and appeal costs are separate.", "comparator": "Narrative baseline is the student paying the fine; no formal comparative design.", "ai_attribution": "AI drafted the appeal; council decision depended on a valid permit and other facts.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.bbc.co.uk/news/uk-england-york-north-yorkshire-65126772", "title": "York student has a £60 parking fine withdrawn after sending an appeal letter drafted by ChatGPT", "quote": "A student has successfully appealed against a £60 parking fine by using a letter written by an artificial intelligence chatbot.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Council decision and date", "AI contribution relative to entitlement", "Time and other appeal costs" ], "overlap_flags": [ "beneficiary_specific_transfer", "single_case", "counterfactual_self_report" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "This is a concrete personal financial benefit; it is not automatically a net social gain of the same amount.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "personal liability or fine avoided", "claim_status": "reported_liability_withdrawal", "currency": "GBP", "currency_basis": "symbol_and_UK_context", "unit_basis": "one parking fine", "time_basis": "Case reported March 2023; exact withdrawal date absent.", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "Narrative baseline is the student paying the fine; no formal comparative design.", "ai_and_value_scope": "AI drafted the appeal; council decision depended on a valid permit and other facts.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross personal payment avoided; transfer effect for the council and appeal costs are separate.", "net_benefit_established": false, "comparability_flags": [ "beneficiary_specific_transfer", "single_case", "counterfactual_self_report" ], "data_gaps": [ "Council decision and date", "AI contribution relative to entitlement", "Time and other appeal costs" ], "extraction_note": "This is a concrete personal financial benefit; it is not automatically a net social gain of the same amount.", "source_review_required": true, "id": "AIBD-0116", "title": "York student has a £60 parking fine withdrawn after sending an appeal letter drafted by ChatGPT", "value": { "amount": "£60", "what": "parking fine withdrawn by York City Council after the ChatGPT-drafted appeal", "who": "Millie Houlton", "estimate": false, "quote": "A student has successfully appealed against a £60 parking fine by using a letter written by an artificial intelligence chatbot.", "url": "https://www.bbc.co.uk/news/uk-england-york-north-yorkshire-65126772" }, "sourceURL": "https://www.bbc.co.uk/news/uk-england-york-north-yorkshire-65126772", "date_observed_as_stored": "2023-03-31", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0013", "case_id": "AIBD-0338", "title": "IRS applies AI and machine learning to select large partnerships for examination (IR-2023–166, 8 September 2023); two AI models prioritize partnership audits", "summary": "additional tax assessments yearly that IRS management estimates from reassigning 14 classifiers to examinations after the Form 1040 return classification AI model automated classification (an SB/SE individual-return model, not the partnership models)", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "approximately $26.1 million", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "USD", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "revenue", "amount_basis": "unresolved", "period": "Annual projection; start and forecast horizon unstated.", "unit": "per year after reassigning 14 classifiers", "beneficiary": "the Internal Revenue Service (U.S. Treasury)", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Projected gross assessments, not collected cash or net return.", "comparator": "No direct monetary comparator or realized collection evidence.", "ai_attribution": "The value concerns Form 1040 individual-return classification, while the record concerns partnership models.", "ai_attribution_status": "claim_link_mismatch", "net_benefit_established": false, "sources": [ { "url": "https://www.oversight.gov/sites/default/files/documents/reports/2025-05/2025308022fr.pdf", "title": "IRS applies AI and machine learning to select large partnerships for examination (IR-2023–166, 8 September 2023); two AI models prioritize partnership audits", "quote": "According to IRS management, automating the classification function has allowed the IRS to reassign 14 classifiers to the examination function and they estimate the increased examinations could generate approximately $26.1 million in additional tax assessments yearly.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Link the claim to the correct model and case", "Projection method", "Actual examinations, assessments and receipts", "Program costs and forecast horizon" ], "overlap_flags": [ "claim_case_scope_mismatch", "assessed_not_collected", "projection", "labor_redeployment_mechanism" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "mismatch", "review_note": "Hold: the projected $26.1 million belongs to Form 1040 staff reassignment, not the partnership-model intervention in this AIBD case. A correct claim-to-case link is required.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "revenue", "money_subtype": "projected additional tax assessments from staff reassignment", "claim_status": "projected", "currency": "USD", "currency_basis": "inferred_from_US_program", "unit_basis": "per year after reassigning 14 classifiers", "time_basis": "Annual projection; start and forecast horizon unstated.", "time_basis_completeness": "annualized_without_window", "monetary_comparator_assessment": "No direct monetary comparator or realized collection evidence.", "ai_and_value_scope": "The value concerns Form 1040 individual-return classification, while the record concerns partnership models.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Projected gross assessments, not collected cash or net return.", "net_benefit_established": false, "comparability_flags": [ "claim_case_scope_mismatch", "assessed_not_collected", "projection", "labor_redeployment_mechanism" ], "data_gaps": [ "Link the claim to the correct model and case", "Projection method", "Actual examinations, assessments and receipts", "Program costs and forecast horizon" ], "extraction_note": "The amount should be a separate claim or case before it is treated as evidence about the partnership models.", "source_review_required": true, "id": "AIBD-0338", "title": "IRS applies AI and machine learning to select large partnerships for examination (IR-2023–166, 8 September 2023); two AI models prioritize partnership audits", "value": { "amount": "approximately $26.1 million", "what": "additional tax assessments yearly that IRS management estimates from reassigning 14 classifiers to examinations after the Form 1040 return classification AI model automated classification (an SB/SE individual-return model, not the partnership models)", "who": "the Internal Revenue Service (U.S. Treasury)", "estimate": true, "quote": "According to IRS management, automating the classification function has allowed the IRS to reassign 14 classifiers to the examination function and they estimate the increased examinations could generate approximately $26.1 million in additional tax assessments yearly.", "url": "https://www.oversight.gov/sites/default/files/documents/reports/2025-05/2025308022fr.pdf" }, "sourceURL": "https://www.oversight.gov/sites/default/files/documents/reports/2025-05/2025308022fr.pdf", "date_observed_as_stored": "2023-09-08", "existing_estimate_flag": true, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0014", "case_id": "AIBD-0013", "title": "Vesuvius Challenge grand prize", "summary": "The competition awarded $700,000 to the three-person team that used machine learning to read a carbonized scroll.", "monetary_role": "recipient_transfer", "amount": { "source_text": "US$700,000", "value": "700000", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "context", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "recipient_transfer", "amount_basis": "transfer", "period": "Grand-prize announcement on February 5, 2024.", "unit": "One prize shared by a three-person team.", "beneficiary": "Youssef Nader, Luke Farritor, and Julian Schilliger.", "payer": "Vesuvius Challenge prize fund.", "realization": "awarded", "accounting_boundary": "Gross prize award to recipients, not new aggregate wealth or research profit.", "comparator": "No cash counterfactual is stated.", "ai_attribution": "Machine learning supported the winning scroll-reading work. The award is a financial consequence of that work, without an isolated AI share.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://scrollprize.org/grandprize", "title": "Vesuvius Challenge 2023 Grand Prize awarded", "quote": "the Vesuvius Challenge Grand Prize of $700,000 is awarded to a team of three", "date": "2024-02-05", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": null } ], "limitations": [ "The announcement establishes an award, not completed payment or the winners’ net proceeds." ], "overlap_flags": [ "One shared prize, not $700,000 per team member." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "recipient_transfer", "money_type": "other", "money_subtype": "prize award to winning researchers", "claim_status": "awarded", "currency": "USD", "currency_basis": "inferred_from_competition_context", "unit_basis": "one team award shared among three winners", "time_basis": "Award announced February 5, 2024", "time_basis_completeness": "dated_one_off", "monetary_comparator_assessment": "The comparator measures scroll text recovered, not financial gain.", "ai_and_value_scope": "Prize awarded for AI-assisted work; award is not a valuation of recovered knowledge.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross award to recipients; research costs and payer outlay are separate.", "net_benefit_established": false, "comparability_flags": [ "recipient_transfer", "prize_not_research_value", "gross_award" ], "data_gaps": [ "Payment confirmation and recipient split if relevant", "Research costs", "Keep award distinct from monetized cultural value" ], "extraction_note": "Potentially useful in a recipient-benefit or research-incentive view, but not a quantified improvement in the cost or value of reading the scroll.", "source_review_required": true, "id": "AIBD-0013", "title": "Vesuvius Challenge 2023: machine-learning ink detection reads over 2,000 characters from unopened Herculaneum scroll PHerc.Paris.4", "value": { "amount": "$700,000", "what": "Vesuvius Challenge grand prize awarded for the winning submission that read the scroll", "who": "Youssef Nader, Luke Farritor and Julian Schilliger, paid by the Vesuvius Challenge", "estimate": false, "quote": "The results of this review were clear and unanimous: the Vesuvius Challenge Grand Prize of $700,000 is awarded to a team of three for their excellent submission. Congratulations to Youssef Nader, Luke Farritor, and Julian Schilliger!", "url": "https://scrollprize.org/grandprize" }, "sourceURL": "https://scrollprize.org/grandprize", "date_observed_as_stored": "2024-02-05", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0015", "case_id": "AIBD-0335", "title": "France's Foncier innovant aerial-image AI detects 140,000 undeclared pools and about €40 million in extra property tax by 20 March 2024; Dordogne alone 1,500 in 2023", "summary": "additional property tax (taxe foncière) to be collected on the 140,000 detected pools", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "40 millions d’euros", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "EUR", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "revenue", "amount_basis": "unresolved", "period": "Reported March 2024; accounts disagree on the annual detection split; some describe 2023 tax.", "unit": "tax from 140,000 detected pools", "beneficiary": "local authorities (collectivités), which receive the taxe foncière", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Expected gross tax revenue; deployment and verification costs are not deducted.", "comparator": "Narrative comparison with manual surveying; no comparable revenue baseline.", "ai_attribution": "Minister attributes the figure to a tool plus human verification.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://rmc.bfmtv.com/actualites/economie/grace-a-l-ia-l-etat-decouvre-140-000-piscines-non-declarees-et-recupere-40-millions-d-euros_AV-202403210525.html", "title": "France's Foncier innovant aerial-image AI detects 140,000 undeclared pools and about €40 million in extra property tax by 20 March 2024; Dordogne alone 1,500 in 2023", "quote": "ce sont 140.000 piscines qui ont été détectées et qui permettront à nos collectivités de récolter 40 millions d’euros au titre de la taxe foncière", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Official tax authority figures", "Tax period and collection status", "Cost of verification and false positives", "No-tool revenue counterfactual" ], "overlap_flags": [ "assessed_not_collected", "period_disagreement", "operator_claim", "human_verification_cost_missing" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "Keep the expected revenue separate from actual collected revenue.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "revenue", "money_subtype": "additional property tax expected", "claim_status": "projected_or_assessed_not_confirmed_collected", "currency": "EUR", "currency_basis": "explicit", "unit_basis": "tax from 140,000 detected pools", "time_basis": "Reported March 2024; accounts disagree on the annual detection split; some describe 2023 tax.", "time_basis_completeness": "partial_window", "monetary_comparator_assessment": "Narrative comparison with manual surveying; no comparable revenue baseline.", "ai_and_value_scope": "Minister attributes the figure to a tool plus human verification.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Expected gross tax revenue; deployment and verification costs are not deducted.", "net_benefit_established": false, "comparability_flags": [ "assessed_not_collected", "period_disagreement", "operator_claim", "human_verification_cost_missing" ], "data_gaps": [ "Official tax authority figures", "Tax period and collection status", "Cost of verification and false positives", "No-tool revenue counterfactual" ], "extraction_note": "Keep the expected revenue separate from actual collected revenue.", "source_review_required": true, "id": "AIBD-0335", "title": "France's Foncier innovant aerial-image AI detects 140,000 undeclared pools and about €40 million in extra property tax by 20 March 2024; Dordogne alone 1,500 in 2023", "value": { "amount": "40 millions d’euros", "what": "additional property tax (taxe foncière) to be collected on the 140,000 detected pools", "who": "local authorities (collectivités), which receive the taxe foncière", "estimate": true, "quote": "ce sont 140.000 piscines qui ont été détectées et qui permettront à nos collectivités de récolter 40 millions d’euros au titre de la taxe foncière", "url": "https://rmc.bfmtv.com/actualites/economie/grace-a-l-ia-l-etat-decouvre-140-000-piscines-non-declarees-et-recupere-40-millions-d-euros_AV-202403210525.html" }, "sourceURL": "https://rmc.bfmtv.com/actualites/economie/grace-a-l-ia-l-etat-decouvre-140-000-piscines-non-declarees-et-recupere-40-millions-d-euros_AV-202403210525.html", "date_observed_as_stored": "2024-03-20", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0016", "case_id": "AIBD-0232", "title": "Cheshire Police facial recognition matches CCTV to Keenan Hughes; county lines dealers jailed for over five years", "summary": "crack cocaine seized from Hughes (99 wraps, approximate street value)", "monetary_role": "reference_or_input_amount", "amount": { "source_text": "£990", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "GBP", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "reference_amount", "amount_basis": "reference", "period": "May 17, 2024", "unit": "99 wraps seized in one arrest", "beneficiary": "Cheshire Constabulary's Northwich Proactive Policing Team (seized from Keenan Hughes)", "payer": "Not established in the current source review.", "realization": "context", "accounting_boundary": "Contraband valuation, not a cash saving or net social benefit.", "comparator": "Identification-speed comparison, not a monetary benefit comparator.", "ai_attribution": "Face match helped identify a suspect; seizure followed patrol and investigation.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.deeside.com/facial-recognition-technology-helps-to-identify-deeside-drug-dealer", "title": "Cheshire Police facial recognition matches CCTV to Keenan Hughes; county lines dealers jailed for over five years", "quote": "99 individual wraps of crack cocaine, worth approximately £990", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "A defensible monetary benefit endpoint if one is sought", "Valuation basis", "Costs and consequences of intervention" ], "overlap_flags": [ "seized_goods_value_not_benefit", "estimated_street_price", "estimate_flag_too_coarse" ], "curation_status": "context_only", "source_review_status": "context_only", "case_link_status": "unresolved", "review_note": "The £990 is not a quantified benefit to the police or community.", "reviewed_on": null, "legacy_audit": { "monetary_role": "reference_or_input_amount", "money_type": "other", "money_subtype": "street value of seized contraband", "claim_status": "estimated_property_value", "currency": "GBP", "currency_basis": "symbol_and_UK_context", "unit_basis": "99 wraps seized in one arrest", "time_basis": "May 17, 2024", "time_basis_completeness": "dated_one_off", "monetary_comparator_assessment": "Identification-speed comparison, not a monetary benefit comparator.", "ai_and_value_scope": "Face match helped identify a suspect; seizure followed patrol and investigation.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Contraband valuation, not a cash saving or net social benefit.", "net_benefit_established": false, "comparability_flags": [ "seized_goods_value_not_benefit", "estimated_street_price", "estimate_flag_too_coarse" ], "data_gaps": [ "A defensible monetary benefit endpoint if one is sought", "Valuation basis", "Costs and consequences of intervention" ], "extraction_note": "The £990 is not a quantified benefit to the police or community.", "source_review_required": true, "id": "AIBD-0232", "title": "Cheshire Police facial recognition matches CCTV to Keenan Hughes; county lines dealers jailed for over five years", "value": { "amount": "£990", "what": "crack cocaine seized from Hughes (99 wraps, approximate street value)", "who": "Cheshire Constabulary's Northwich Proactive Policing Team (seized from Keenan Hughes)", "estimate": false, "quote": "99 individual wraps of crack cocaine, worth approximately £990", "url": "https://www.deeside.com/facial-recognition-technology-helps-to-identify-deeside-drug-dealer" }, "sourceURL": "https://www.deeside.com/facial-recognition-technology-helps-to-identify-deeside-drug-dealer", "date_observed_as_stored": "2024-05-17", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0017", "case_id": "AIBD-0095", "title": "Flood-triggered cash transfers in Nigeria", "summary": "GiveDirectly reports that it paid $105 per person in an AI-informed flood-response pilot in Kogi, Nigeria.", "monetary_role": "recipient_transfer", "amount": { "source_text": "US$105 per recipient", "value": "105", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "context", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [ { "id": "ngn-equivalent", "amount": { "source_text": "NGN177,866 per recipient", "value": "177866", "lower": null, "upper": null, "qualifier": "exact", "currency": "NGN", "currency_basis": "explicit", "price_basis": "Nominal source amount; price year not separately stated." }, "relationship": "source_equivalent", "equivalent_to": "primary", "equivalence_note": "Reported equivalent; later naira payments may differ.", "period": "August 29 to October 24, 2024, for the first transfer.", "unit": "Per recipient, first transfer.", "realization": "paid", "accounting_boundary": "Recipient cash transfer. A transfer is not a net increase in societal resources or an estimate of avoided flood loss.", "sources": [ { "url": "https://www.aljazeera.com/features/2024/9/21/could-ai-save-nigerians-from-devastating-floods", "title": "Could AI save Nigerians from devastating floods?", "quote": "177,866 naira ($105)", "date": "2024-09-21", "primary": false, "written_by": "independent", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "First anticipatory payment in Kogi" } ], "source_review_status": "source_checked", "review_note": "Denomination checked." } ], "financial_type": "recipient_transfer", "amount_basis": "transfer", "period": "August 29 to October 24, 2024, for the first transfer.", "unit": "Per recipient, first transfer.", "beneficiary": "Registered recipients in six wards of Kogi, Nigeria.", "payer": "GiveDirectly and its funders.", "realization": "paid", "accounting_boundary": "Recipient cash transfer. A transfer is not a net increase in societal resources or an estimate of avoided flood loss.", "comparator": "The amount is not a comparison with aid delivered without AI.", "ai_attribution": "AI forecasts informed targeting and payment triggers alongside local observations and human verification.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://www.givedirectly.org/flood-forecast-ai", "title": "How AI helped send cash just before floods hit in Nigeria", "quote": "we sent $105 to over 4,600 individuals", "date": "2025-06-27", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": null } ], "limitations": [ "The detailed account says that 3,250 people received early payments and another 1,400 were paid after the October 21 peak; the summary overstates early coverage.", "Transfer amount is implementer-reported." ], "overlap_flags": [ "A later $210 transfer is separate. Do not treat the $105 amount as a program total." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "recipient_transfer", "money_type": "other", "money_subtype": "anticipatory cash transfer", "claim_status": "reported_paid_transfer_with_timing_dispute", "currency": "USD", "currency_basis": "currency_inferred_from_implementer_context", "unit_basis": "per recipient; the account reports ambiguity between individual and household", "time_basis": "Transfers August-October 2024; a separate $210 transfer followed.", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "Income and food-security before/after survey; no controlled monetary-impact estimate.", "ai_and_value_scope": "AI contributed to targeting and timing; the cash is the program transfer, not the incremental value of AI.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross recipient transfer and donor outlay; not net social gain.", "net_benefit_established": false, "comparability_flags": [ "recipient_transfer", "intervention_input", "recipient_denominator_ambiguous", "late_payment_subset", "no_control" ], "data_gaps": [ "Individual versus household denominator", "Which payments preceded flooding", "Amount-specific cohort", "AI timing benefit relative to unconditional or later cash", "Program costs" ], "extraction_note": "The $105 can be tracked as money received, while the separate income change is a distinct outcome claim.", "source_review_required": true, "id": "AIBD-0095", "title": "GiveDirectly paid anticipatory cash in Kogi State, Nigeria, on Google Flood Hub forecast triggers (August to October 2024)", "value": { "amount": "$105", "what": "anticipatory cash transfer per recipient before the flood peak (a second $210 transfer followed)", "who": "over 4,600 flood-exposed residents of Ibaji, Kogi State", "estimate": false, "quote": "Once forecasted flood thresholds were met, we sent $105 to over 4,600 individuals before the flood peak.", "url": "https://www.givedirectly.org/flood-forecast-ai/" }, "sourceURL": "https://www.givedirectly.org/flood-forecast-ai/", "date_observed_as_stored": "2024-08-29", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0018", "case_id": "AIBD-0144", "title": "Blackpool road-maintenance savings", "summary": "Blackpool Council reports that technology-assisted repairs cost over £1 million less than its estimate for traditional methods.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "Over £1 million", "value": "1000000", "lower": null, "upper": null, "qualifier": "more_than", "currency": "GBP", "currency_basis": "context", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "incremental_change", "period": "Since summer 2019 to the council account; the endpoint is not specified in the text.", "unit": "5,145 potholes in Blackpool’s North Shore.", "beneficiary": "Blackpool Council.", "payer": "Council road-maintenance budget; savings are avoided expenditure.", "realization": "retrospective_estimate", "accounting_boundary": "Reported repair-cost difference against an estimated traditional-method cost, without a fully costed AI-only comparison.", "comparator": "About £1.5 million under traditional methods versus just under £450,000 reported cost.", "ai_attribution": "AI-assisted condition surveys supported a wider maintenance approach. The source does not isolate AI from the repair and workflow changes.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://blackpoolunlimited.com/artificial-intelligence-boosts-road-maintenance-efficiency/", "title": "Artificial intelligence boosts road maintenance efficiency", "quote": "save over £1m resulting in a final cost of just under £450k.", "date": null, "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "Council account of Project Amber’s North Shore repairs" } ], "limitations": [ "Council-reported costs and a modeled traditional-method baseline." ], "overlap_flags": [ "The two underlying cost figures form this difference and must not be added as separate benefits." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "road-repair expenditure avoided", "claim_status": "reported_saving_against_estimated_traditional_cost", "currency": "GBP", "currency_basis": "symbol_and_UK_context", "unit_basis": "5,145 potholes in North Shore", "time_basis": "Since summer 2019; exact end of repair cohort not stated on the cited value page.", "time_basis_completeness": "incomplete_window", "monetary_comparator_assessment": "Value/account compare about £1.5m traditional cost with under £450k actual cost; the structured comparator instead concerns compensation payouts.", "ai_and_value_scope": "The entire Project Amber process combines imagery, prioritization and surface treatment.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross repair saving relative to a traditional-cost estimate; total technology/program costs are unspecified.", "net_benefit_established": false, "comparability_flags": [ "whole_program_attribution", "comparator_field_different_outcome", "estimated_counterfactual", "cohort_end_unclear" ], "data_gaps": [ "Repair-cohort period", "Basis of traditional-cost estimate", "Like-for-like repair quality and scope", "All implementation and ongoing costs" ], "extraction_note": "Do not combine repair savings with reduced compensation payouts without distinct claims and overlap checks.", "source_review_required": true, "id": "AIBD-0144", "title": "Blackpool Council's Project Amber imagery and AI road surveys: 5,145 potholes filled for about £450,000 and pothole claim payouts down to £719", "value": { "amount": "over £1m", "what": "saved on filling 5,145 potholes in North Shore, final cost just under £450k against £1.5m by traditional methods", "who": "Blackpool Council", "estimate": false, "quote": "Blackpool Council has managed to save over £1m resulting in a final cost of just under £450k.", "url": "https://blackpoolunlimited.com/artificial-intelligence-boosts-road-maintenance-efficiency/" }, "sourceURL": "https://blackpoolunlimited.com/artificial-intelligence-boosts-road-maintenance-efficiency/", "date_observed_as_stored": "2024-09-26", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0019", "case_id": "AIBD-0173", "title": "Treasury check-fraud recoveries", "summary": "Treasury reports that machine learning helped identify check fraud associated with $1 billion recovered in fiscal 2024.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "US$1 billion", "value": "1000000000", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "context", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "recovered_funds", "amount_basis": "incremental_change", "period": "U.S. fiscal year 2024.", "unit": "Treasury check-fraud recoveries across the program.", "beneficiary": "U.S. Treasury and affected federal programs.", "payer": "Recovered fraud proceeds; individual payers are not identified.", "realization": "reported_recovery", "accounting_boundary": "Gross recovered funds. Detection, recovery, and operating costs are not deducted.", "comparator": "The release does not estimate recoveries without machine learning.", "ai_attribution": "Treasury states that machine learning expedited check-fraud identification. The amount is not an estimate of the incremental gain from AI alone.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://home.treasury.gov/news/press-releases/jy2650", "title": "Treasury announces enhanced fraud detection processes, including machine learning AI", "quote": "Expediting the identification of Treasury check fraud with machine learning AI resulting in $1 billion in recovery.", "date": "2024-10-17", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": null } ], "limitations": [ "Agency-reported recovery, without a counterfactual AI-only estimate or full program costs." ], "overlap_flags": [ "Component of Treasury’s broader fiscal-2024 recovery and prevention total; do not add it to that total." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "recovered check fraud", "claim_status": "reported_recovery", "currency": "USD", "currency_basis": "inferred_from_US_program", "unit_basis": "annual machine-learning check-fraud recovery", "time_basis": "FY2024: October 2023-September 2024", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "Structured comparator compares the full prevention/recovery program across years, not the $1bn AI component.", "ai_and_value_scope": "Treasury explicitly attributes this component to machine-learning identification; incremental causality remains unevaluated.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross recovered funds; recovery and system costs unspecified.", "net_benefit_established": false, "comparability_flags": [ "component_vs_program_comparator", "recovered_not_prevented", "operator_claim", "overlap_with_4_billion_total" ], "data_gaps": [ "Recovery accounting and audit", "Program costs", "Equivalent prior-year ML recovery", "Incremental contribution relative to investigators" ], "extraction_note": "The $1bn is a component of the greater-than-$4bn total, not an amount to add to that total.", "source_review_required": true, "id": "AIBD-0173", "title": "Treasury's machine-learning check-fraud detection recovers $1 billion in FY2024", "value": { "amount": "$1 billion", "what": "Treasury check fraud recovered in FY2024 through machine-learning identification", "who": "U.S. Treasury, on behalf of taxpayers", "estimate": false, "quote": "Expediting the identification of Treasury check fraud with machine learning AI resulting in $1 billion in recovery.", "url": "https://home.treasury.gov/news/press-releases/jy2650" }, "sourceURL": "https://home.treasury.gov/news/press-releases/jy2650", "date_observed_as_stored": "2024-10-17", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0020", "case_id": "AIBD-0336", "title": "Udbetaling Danmark's fraud-control models may have produced 135, 36 and 72 cases with benefits stopped or reduced in 2023", "summary": "erroneous payments unveiled by UDK and municipal controls in 2017 and 2018; the source says it is unclear how much was recovered and that the figure includes averted future payments, and it does not isolate the models", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "half a billion Danish crowns", "value": "500000000", "lower": null, "upper": null, "qualifier": "exact", "currency": "DKK", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [ { "id": "eur-equivalent", "amount": { "source_text": "approximately 70 million euros", "value": "70000000", "lower": null, "upper": null, "qualifier": "approximately", "currency": "EUR", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "relationship": "source_equivalent", "equivalent_to": "primary", "equivalence_note": "The source gives both currencies for the same amount. It does not specify an exchange-rate date or method.", "period": "2017 and 2018; the main case describes model outcomes in 2023.", "unit": "UDK and municipal control aggregate", "realization": "unverified", "accounting_boundary": "Mixture of recovery, unrecoverable amounts and future avoidance; net meaning unknown.", "sources": [ { "url": "https://algorithmwatch.org/en/udbetaling-danmark/", "title": "In a quest to optimize welfare management, Denmark built a surveillance behemoth", "quote": "half a billion Danish crowns in erroneous payment (approximately 70 million euros)", "date": "2020-08-06", "primary": false, "written_by": "independent", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "91 referrals to the police" } ], "source_review_status": "source_checked", "review_note": "The currency expression is explicit in the source. Its equivalence does not establish a separate gain or resolve the claim’s remaining evidence gaps." } ], "financial_type": "avoided_loss", "amount_basis": "unresolved", "period": "2017 and 2018; the main case describes model outcomes in 2023.", "unit": "UDK and municipal control aggregate", "beneficiary": "Udbetaling Danmark and the municipalities (the Danish public purse)", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Mixture of recovery, unrecoverable amounts and future avoidance; net meaning unknown.", "comparator": "No comparator.", "ai_attribution": "Historic all-control amount is not isolated to the named models or current outcome period.", "ai_attribution_status": "claim_link_mismatch", "net_benefit_established": false, "sources": [ { "url": "https://algorithmwatch.org/en/udbetaling-danmark/", "title": "In a quest to optimize welfare management, Denmark built a surveillance behemoth", "quote": "In 2017 and 2018, controls by UDK and municipalities unveiled half a billion Danish crowns in erroneous payment (approximately 70 million euros). It is unclear what amount was recovered, what was not recovered due to the beneficiary’s incapacity to repay, and what was an accounting trick (the figure includes future payments that were averted because of the control).", "date": "2020-08-06", "primary": false, "written_by": "independent", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "91 referrals to the police" } ], "limitations": [ "Recovered cash versus future avoided payments versus identified debt", "AI-model contribution", "Correct claim period and cohort", "Costs and claimant consequences" ], "overlap_flags": [ "mixed_accounting", "claim_case_period_mismatch", "whole_control_program", "dual_currency_same_amount", "estimate_flag_too_coarse" ], "curation_status": "awaiting_review", "source_review_status": "source_checked_link_unresolved", "case_link_status": "mismatch", "review_note": "The half-billion-crown amount concerns 2017–2018, while the linked model outcomes concern 2023. Recovered funds, future avoided payments, and the models' contribution remain unresolved.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "erroneous payments identified, recovered or averted", "claim_status": "mixed_and_accounting_unresolved", "currency": "DKK; EUR equivalent", "currency_basis": "explicit", "unit_basis": "UDK and municipal control aggregate", "time_basis": "2017 and 2018; the main case describes model outcomes in 2023.", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "No comparator.", "ai_and_value_scope": "Historic all-control amount is not isolated to the named models or current outcome period.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Mixture of recovery, unrecoverable amounts and future avoidance; net meaning unknown.", "net_benefit_established": false, "comparability_flags": [ "mixed_accounting", "claim_case_period_mismatch", "whole_control_program", "dual_currency_same_amount", "estimate_flag_too_coarse" ], "data_gaps": [ "Recovered cash versus future avoided payments versus identified debt", "AI-model contribution", "Correct claim period and cohort", "Costs and claimant consequences" ], "extraction_note": "Do not present this historic aggregate as revenue generated by the three models in 2023.", "source_review_required": true, "id": "AIBD-0336", "title": "Udbetaling Danmark's fraud-control models may have produced 135, 36 and 72 cases with benefits stopped or reduced in 2023", "value": { "amount": "half a billion Danish crowns (approximately 70 million euros)", "what": "erroneous payments unveiled by UDK and municipal controls in 2017 and 2018; the source says it is unclear how much was recovered and that the figure includes averted future payments, and it does not isolate the models", "who": "Udbetaling Danmark and the municipalities (the Danish public purse)", "estimate": false, "quote": "In 2017 and 2018, controls by UDK and municipalities unveiled half a billion Danish crowns in erroneous payment (approximately 70 million euros). It is unclear what amount was recovered, what was not recovered due to the beneficiary’s incapacity to repay, and what was an accounting trick (the figure includes future payments that were averted because of the control).", "url": "https://algorithmwatch.org/en/udbetaling-danmark/" }, "sourceURL": "https://algorithmwatch.org/en/udbetaling-danmark/", "date_observed_as_stored": "2024-11-12", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0021", "case_id": "AIBD-0087", "title": "CGU ALICE alert leads to a cut of about R$100 million in the reference value of a federal road-works tender (30 December 2024)", "summary": "reduction in the reference value of a federal road-works engineering tender after the ALICE alert", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "cerca de R$ 100 milhões", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "BRL", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "unresolved", "period": "Reported December 2024; actual case date is not established.", "unit": "one road-works engineering tender", "beneficiary": "the federal government (contracting agency not named)", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross reduction of a reference budget; actual final expenditure and total costs unknown.", "comparator": "Original reference price exists but is not quantified; reported delta is about R$100m.", "ai_attribution": "ALICE alert led to human audit and management adjustment.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://web.archive.org/web/20260207114437id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2024/12/orgaos-federais-economizaram-em-2024-257-milhoes-de-reais-com-apoio-do-robo-alice-da-controladoria-geral-da-uniao", "title": "CGU ALICE alert leads to a cut of about R$100 million in the reference value of a federal road-works tender (30 December 2024)", "quote": "Outro exemplo relevante dos benefícios decorrentes do uso da ferramenta foi a redução de cerca de R$ 100 milhões no valor de referência em um edital para contratação de serviços de engenharia.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Original and revised prices", "Final award/payment and equivalent scope", "Case date and contracting body", "Overlap with annual R$257.25m aggregate" ], "overlap_flags": [ "budget_revision_not_cash_release", "approximate_amount", "possible_overlap_with_2024_aggregate", "event_date_missing" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "This is a monetary change, unlike a tender face value, but should be labeled reference-price cost avoidance.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "procurement reference-price reduction", "claim_status": "reported_budget_reduction", "currency": "BRL", "currency_basis": "explicit", "unit_basis": "one road-works engineering tender", "time_basis": "Reported December 2024; actual case date is not established.", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "Original reference price exists but is not quantified; reported delta is about R$100m.", "ai_and_value_scope": "ALICE alert led to human audit and management adjustment.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross reduction of a reference budget; actual final expenditure and total costs unknown.", "net_benefit_established": false, "comparability_flags": [ "budget_revision_not_cash_release", "approximate_amount", "possible_overlap_with_2024_aggregate", "event_date_missing" ], "data_gaps": [ "Original and revised prices", "Final award/payment and equivalent scope", "Case date and contracting body", "Overlap with annual R$257.25m aggregate" ], "extraction_note": "This is a monetary change, unlike a tender face value, but should be labeled reference-price cost avoidance.", "source_review_required": true, "id": "AIBD-0087", "title": "CGU ALICE alert leads to a cut of about R$100 million in the reference value of a federal road-works tender (30 December 2024)", "value": { "amount": "cerca de R$ 100 milhões", "what": "reduction in the reference value of a federal road-works engineering tender after the ALICE alert", "who": "the federal government (contracting agency not named)", "estimate": false, "quote": "Outro exemplo relevante dos benefícios decorrentes do uso da ferramenta foi a redução de cerca de R$ 100 milhões no valor de referência em um edital para contratação de serviços de engenharia.", "url": "https://web.archive.org/web/20260207114437id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2024/12/orgaos-federais-economizaram-em-2024-257-milhoes-de-reais-com-apoio-do-robo-alice-da-controladoria-geral-da-uniao" }, "sourceURL": "https://web.archive.org/web/20260207114437id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2024/12/orgaos-federais-economizaram-em-2024-257-milhoes-de-reais-com-apoio-do-robo-alice-da-controladoria-geral-da-uniao", "date_observed_as_stored": "2024-12-30", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0022", "case_id": "AIBD-0081", "title": "CGU's ALICE robot flags overpricing in three 2024 Rio de Janeiro federal hospital tenders; one suspended, one revoked, one re-estimated", "summary": "reduction of the estimated value of tender 90017/2024 after the re-estimate (the release gives no total for the three tenders)", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "R$ 3.063.774,84", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "BRL", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "unresolved", "period": "2024 tender; result reported January 14, 2025; precise revision date absent.", "unit": "tender 90017/2024 at Hospital Federal de Ipanema", "beneficiary": "Hospital Federal de Ipanema, federal budget", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross reduction of an estimate; actual paid saving and audit/system costs unknown.", "comparator": "Explicit original R$15,450,714.80 versus revised R$12,386,939.96.", "ai_attribution": "ALICE alert followed by auditors and hospital managers.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://web.archive.org/web/20250114172015id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2024/12/atuacao-do-robo-alice-segue-auxiliando-auditores-a-realizar-trabalhos-preventivos-que-ajudam-a-evitar-desperdicios-e-irregularidades/", "title": "CGU's ALICE robot flags overpricing in three 2024 Rio de Janeiro federal hospital tenders; one suspended, one revoked, one re-estimated", "quote": "Como resultado da atuação, os gestores procederam nova estimativa de preços, reduzindo o valor total estimado da compra para R$ 12.386.939,96, o que representou uma economia de mais de R$ 3 milhões (R$ 3.063.774,84).", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Final contract and payment", "Equivalent quantity and quality", "Revision date", "Avoid overlap with broader audits or annual totals" ], "overlap_flags": [ "budget_revision_not_cash_release", "single_component_of_three_cases", "possible_overlap_with_program_aggregates" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "The cited R$3,063,774.84 is a documented estimate change, not the sum of all three tenders or all flagged overpricing risks.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "procurement estimate reduction", "claim_status": "reported_budget_reduction", "currency": "BRL", "currency_basis": "explicit", "unit_basis": "tender 90017/2024 at Hospital Federal de Ipanema", "time_basis": "2024 tender; result reported January 14, 2025; precise revision date absent.", "time_basis_completeness": "partial_window", "monetary_comparator_assessment": "Explicit original R$15,450,714.80 versus revised R$12,386,939.96.", "ai_and_value_scope": "ALICE alert followed by auditors and hospital managers.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross reduction of an estimate; actual paid saving and audit/system costs unknown.", "net_benefit_established": false, "comparability_flags": [ "budget_revision_not_cash_release", "single_component_of_three_cases", "possible_overlap_with_program_aggregates" ], "data_gaps": [ "Final contract and payment", "Equivalent quantity and quality", "Revision date", "Avoid overlap with broader audits or annual totals" ], "extraction_note": "The cited R$3,063,774.84 is a documented estimate change, not the sum of all three tenders or all flagged overpricing risks.", "source_review_required": true, "id": "AIBD-0081", "title": "CGU's ALICE robot flags overpricing in three 2024 Rio de Janeiro federal hospital tenders; one suspended, one revoked, one re-estimated", "value": { "amount": "R$ 3.063.774,84", "what": "reduction of the estimated value of tender 90017/2024 after the re-estimate (the release gives no total for the three tenders)", "who": "Hospital Federal de Ipanema, federal budget", "estimate": false, "quote": "Como resultado da atuação, os gestores procederam nova estimativa de preços, reduzindo o valor total estimado da compra para R$ 12.386.939,96, o que representou uma economia de mais de R$ 3 milhões (R$ 3.063.774,84).", "url": "https://web.archive.org/web/20250114172015id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2024/12/atuacao-do-robo-alice-segue-auxiliando-auditores-a-realizar-trabalhos-preventivos-que-ajudam-a-evitar-desperdicios-e-irregularidades/" }, "sourceURL": "https://web.archive.org/web/20250114172015id_/https://www.gov.br/cgu/pt-br/assuntos/noticias/2024/12/atuacao-do-robo-alice-segue-auxiliando-auditores-a-realizar-trabalhos-preventivos-que-ajudam-a-evitar-desperdicios-e-irregularidades/", "date_observed_as_stored": "2025-01-14", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0023", "case_id": "AIBD-0126", "title": "Delhi Police matched a CCTV frame to a prior arrest record with Israeli facial recognition software, arrested two men within about a day of a Rs 80 lakh armed robbery in Chandni Chowk and recovered Rs 79.5 lakh", "summary": "stolen cash recovered", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "Rs 79.5 lakh", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "INR", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "avoided_loss", "amount_basis": "unresolved", "period": "Robbery March 17, 2025; recovery by March 19, 2025", "unit": "one robbery and cash seizure", "beneficiary": "the robbed cash courier's firm, R K Enterprises (recovered by Delhi Police from the accused's residence)", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross cash recovered; return to the victim and investigation costs are not documented.", "comparator": "Rs 80 lakh stolen and Rs 79.5 lakh recovered provide a recovery denominator; there is no no-AI counterfactual.", "ai_attribution": "Face matching was one investigative lead; conventional police work produced recovery.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.rediff.com/news/report/israeli-tech-helps-delhi-police-crack-rs-80-lakh-robbery-2-held/20250319.htm", "title": "Delhi Police matched a CCTV frame to a prior arrest record with Israeli facial recognition software, arrested two men within about a day of a Rs 80 lakh armed robbery in Chandni Chowk and recovered Rs 79.5 lakh", "quote": "A raid at Samir's residence led to the recovery of Rs 79.5 lakh in cash, a semi-automatic pistol with three bullets, and other incriminating materials.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Cash returned to owner versus held as evidence", "Investigative costs", "AI incremental contribution" ], "overlap_flags": [ "recovery_not_expected_loss", "human_investigation_chain", "restoration_not_new_production" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "Rs 79.5 lakh is the recovery amount; Rs 80 lakh is the robbery total, not another benefit.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "recovered stolen cash", "claim_status": "reported_recovery", "currency": "INR", "currency_basis": "inferred_from_Indian_rupees_context", "unit_basis": "one robbery and cash seizure", "time_basis": "Robbery March 17, 2025; recovery by March 19, 2025", "time_basis_completeness": "dated_one_off", "monetary_comparator_assessment": "Rs 80 lakh stolen and Rs 79.5 lakh recovered provide a recovery denominator; there is no no-AI counterfactual.", "ai_and_value_scope": "Face matching was one investigative lead; conventional police work produced recovery.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross cash recovered; return to the victim and investigation costs are not documented.", "net_benefit_established": false, "comparability_flags": [ "recovery_not_expected_loss", "human_investigation_chain", "restoration_not_new_production" ], "data_gaps": [ "Cash returned to owner versus held as evidence", "Investigative costs", "AI incremental contribution" ], "extraction_note": "Rs 79.5 lakh is the recovery amount; Rs 80 lakh is the robbery total, not another benefit.", "source_review_required": true, "id": "AIBD-0126", "title": "Delhi Police matched a CCTV frame to a prior arrest record with Israeli facial recognition software, arrested two men within about a day of a Rs 80 lakh armed robbery in Chandni Chowk and recovered Rs 79.5 lakh", "value": { "amount": "Rs 79.5 lakh", "what": "stolen cash recovered", "who": "the robbed cash courier's firm, R K Enterprises (recovered by Delhi Police from the accused's residence)", "estimate": false, "quote": "A raid at Samir's residence led to the recovery of Rs 79.5 lakh in cash, a semi-automatic pistol with three bullets, and other incriminating materials.", "url": "https://www.rediff.com/news/report/israeli-tech-helps-delhi-police-crack-rs-80-lakh-robbery-2-held/20250319.htm" }, "sourceURL": "https://www.rediff.com/news/report/israeli-tech-helps-delhi-police-crack-rs-80-lakh-robbery-2-held/20250319.htm", "date_observed_as_stored": "2025-03-19", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0024", "case_id": "AIBD-0145", "title": "HMRC's Connect data-matching system: about £4.6 billion of additional yield in 2024–25", "summary": "additional tax yield from Connect cases in the 2024-25 tax year (average £3.4bn a year)", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "£4.6bn", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "GBP", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "revenue", "amount_basis": "unresolved", "period": "2024-25 tax year", "unit": "annual yield from Connect cases", "beneficiary": "HMRC, for the Exchequer, gains", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross additional yield is suggested; assessment, actual cash and future-benefit components are not separated.", "comparator": "No monetary comparator beyond an unstated multi-year average.", "ai_attribution": "Yield attributed to Connect-supported cases; existing sources describe analytics/risk rules without establishing machine learning.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.pinsentmasons.com/out-law/news/hmrc-big-data-tax-revenue-boost-4-6-billion", "title": "HMRC's Connect data-matching system: about £4.6 billion of additional yield in 2024–25", "quote": "HMRC, in a freedom of information request to Pinsent Masons, said it generated on average £3.4bn in additional annual yield from Connect cases, with the 2024-25 tax year generating approximately £4.6bn.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "FOI response and yield definition", "Actual receipts versus assessments or future revenue", "Identifiable AI component", "System and caseworker costs" ], "overlap_flags": [ "yield_definition_missing", "ai_component_needs_review", "whole_program_attribution", "average_not_additive" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "Do not add the £3.4bn average annual figure to the specific-year £4.6bn figure.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "revenue", "money_subtype": "additional tax yield", "claim_status": "reported_yield_with_accounting_definition_missing", "currency": "GBP", "currency_basis": "symbol_and_UK_context", "unit_basis": "annual yield from Connect cases", "time_basis": "2024-25 tax year", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "No monetary comparator beyond an unstated multi-year average.", "ai_and_value_scope": "Yield attributed to Connect-supported cases; existing sources describe analytics/risk rules without establishing machine learning.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross additional yield is suggested; assessment, actual cash and future-benefit components are not separated.", "net_benefit_established": false, "comparability_flags": [ "yield_definition_missing", "ai_component_needs_review", "whole_program_attribution", "average_not_additive" ], "data_gaps": [ "FOI response and yield definition", "Actual receipts versus assessments or future revenue", "Identifiable AI component", "System and caseworker costs" ], "extraction_note": "Do not add the £3.4bn average annual figure to the specific-year £4.6bn figure.", "source_review_required": true, "id": "AIBD-0145", "title": "HMRC's Connect data-matching system: about £4.6 billion of additional yield in 2024–25", "value": { "amount": "£4.6bn", "what": "additional tax yield from Connect cases in the 2024-25 tax year (average £3.4bn a year)", "who": "HMRC, for the Exchequer, gains", "estimate": false, "quote": "HMRC, in a freedom of information request to Pinsent Masons, said it generated on average £3.4bn in additional annual yield from Connect cases, with the 2024-25 tax year generating approximately £4.6bn.", "url": "https://www.pinsentmasons.com/out-law/news/hmrc-big-data-tax-revenue-boost-4-6-billion" }, "sourceURL": "https://www.pinsentmasons.com/out-law/news/hmrc-big-data-tax-revenue-boost-4-6-billion", "date_observed_as_stored": "2025-03-31", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0025", "case_id": "AIBD-0129", "title": "Taoyuan's AI sound cameras locate a white BMW that had been roaring down Jieshou Road at night, within a day of residents' complaint", "summary": "maximum fine under the Noise Control Act to be imposed for the noise violation", "monetary_role": "reference_or_input_amount", "amount": { "source_text": "3,600元 (New Taiwan dollars)", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "TWD", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "reference_amount", "amount_basis": "reference", "period": "Violation detected April 23, 2025; future fine mentioned.", "unit": "one noise violation; repeat penalties possible", "beneficiary": "the BMW's owner pays", "payer": "Not established in the current source review.", "realization": "context", "accounting_boundary": "Legal ceiling, not realized gross receipt or monetized community benefit.", "comparator": "Identification-speed comparison only.", "ai_attribution": "AI-assisted detection may lead to a fine; no assessed or paid amount is established.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://n.yam.com/Article/20250429841522", "title": "Taoyuan's AI sound cameras locate a white BMW that had been roaring down Jieshou Road at night, within a day of residents' complaint", "quote": "後續將依噪音管制法進行裁罰，最高可處3,600元罰鍰，並通知限期改善，屆期未改善者將按次處罰直至符合機動車輛噪音管制標準為止。", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Actual assessed penalty", "Collection status", "Measured noise or welfare improvement" ], "overlap_flags": [ "maximum_not_expected_value", "possible_penalty_not_outcome_amount", "transfer_perspective" ], "curation_status": "context_only", "source_review_status": "context_only", "case_link_status": "unresolved", "review_note": "Do not treat the legal maximum as a projected monetary benefit unless a source actually estimates that outcome.", "reviewed_on": null, "legacy_audit": { "monetary_role": "reference_or_input_amount", "money_type": "other", "money_subtype": "statutory maximum penalty", "claim_status": "possible_penalty_ceiling", "currency": "TWD", "currency_basis": "explicit", "unit_basis": "one noise violation; repeat penalties possible", "time_basis": "Violation detected April 23, 2025; future fine mentioned.", "time_basis_completeness": "dated_one_off", "monetary_comparator_assessment": "Identification-speed comparison only.", "ai_and_value_scope": "AI-assisted detection may lead to a fine; no assessed or paid amount is established.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Legal ceiling, not realized gross receipt or monetized community benefit.", "net_benefit_established": false, "comparability_flags": [ "maximum_not_expected_value", "possible_penalty_not_outcome_amount", "transfer_perspective" ], "data_gaps": [ "Actual assessed penalty", "Collection status", "Measured noise or welfare improvement" ], "extraction_note": "Do not treat the legal maximum as a projected monetary benefit unless a source actually estimates that outcome.", "source_review_required": true, "id": "AIBD-0129", "title": "Taoyuan's AI sound cameras locate a white BMW that had been roaring down Jieshou Road at night, within a day of residents' complaint", "value": { "amount": "3,600元 (New Taiwan dollars)", "what": "maximum fine under the Noise Control Act to be imposed for the noise violation", "who": "the BMW's owner pays", "estimate": true, "quote": "後續將依噪音管制法進行裁罰，最高可處3,600元罰鍰，並通知限期改善，屆期未改善者將按次處罰直至符合機動車輛噪音管制標準為止。", "url": "https://n.yam.com/Article/20250429841522" }, "sourceURL": "https://n.yam.com/Article/20250429841522", "date_observed_as_stored": "2025-04-23", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0026", "case_id": "AIBD-0187", "title": "Medicare payments withheld in Operation Gold Rush", "summary": "DOJ reports that $4.41 billion was held in escrow; the source describes data analytics without establishing that AI was used.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "US$4.41 billion in escrow", "value": "4410000000", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "avoided_loss", "amount_basis": "avoided_expenditure", "period": "Reported June 30, 2025; transaction interval not fully specified.", "unit": "Operation Gold Rush payment block", "beneficiary": "Medicare trust fund (HHS and CMS)", "payer": "Not established in the current source review.", "realization": "withheld_in_escrow", "accounting_boundary": "Scheduled payments withheld in escrow, not cash already returned to the Medicare trust fund.", "comparator": "Scheduled payment about $4.45bn versus about $41m paid, with about $4.41bn escrowed.", "ai_attribution": "Detection is attributed to proactive data analytics; the record states that AI/ML use is unestablished.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.justice.gov/opa/pr/national-health-care-fraud-takedown-results-324-defendants-charged-connection-over-146", "title": "2025 National Health Care Fraud Takedown", "quote": "HHS and CMS intend to seek to return the $4.41 billion in escrow to the Medicare trust fund", "date": "2025-06-30", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": null } ], "limitations": [ "Confirm final escrow disposition", "Identify any actual AI component", "Payment dates and overlap with other fraud aggregates", "Costs", "The source does not identify an AI component in the detection.", "The amount was in escrow, with return to the trust fund still intended." ], "overlap_flags": [ "held_not_recovered", "ai_component_needs_review", "intended_loss_not_paid_loss", "rounded_amounts", "Included in broader Medicare prevented-payment accounts; not additive to those totals." ], "curation_status": "awaiting_review", "source_review_status": "source_checked_link_unresolved", "case_link_status": "matched", "review_note": "The monetary amount is supported, but the cited source does not identify AI in the detection. The future AI-enabled Fusion Center is a separate initiative.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "fraudulent disbursement withheld in escrow", "claim_status": "reported_withheld_with_future_return", "currency": "USD", "currency_basis": "inferred_from_US_program", "unit_basis": "Operation Gold Rush payment block", "time_basis": "Reported June 30, 2025; transaction interval not fully specified.", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "Scheduled payment about $4.45bn versus about $41m paid, with about $4.41bn escrowed.", "ai_and_value_scope": "Detection is attributed to proactive data analytics; the record states that AI/ML use is unestablished.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross funds withheld; eventual return and enforcement costs are separate.", "net_benefit_established": false, "comparability_flags": [ "held_not_recovered", "ai_component_needs_review", "intended_loss_not_paid_loss", "rounded_amounts" ], "data_gaps": [ "Confirm final escrow disposition", "Identify any actual AI component", "Payment dates and overlap with other fraud aggregates", "Costs" ], "extraction_note": "The $10.6bn billed and $14.6bn takedown-wide intended loss must not be added as further benefits.", "source_review_required": true, "id": "AIBD-0187", "title": "DOJ data analytics flag a $10.6 billion catheter billing scheme; CMS withholds $4.41 billion (Operation Gold Rush)", "value": { "amount": "$4.41 billion", "what": "fraudulent Medicare payments held in escrow rather than paid out, to be returned to the Medicare trust fund", "who": "Medicare trust fund (HHS and CMS)", "estimate": false, "quote": "HHS and CMS intend to seek to return the $4.41 billion in escrow to the Medicare trust fund for needed medical care.", "url": "https://www.justice.gov/opa/pr/national-health-care-fraud-takedown-results-324-defendants-charged-connection-over-146" }, "sourceURL": "https://www.justice.gov/opa/pr/national-health-care-fraud-takedown-results-324-defendants-charged-connection-over-146", "date_observed_as_stored": "2025-06-30", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0027", "case_id": "AIBD-0147", "title": "Taichung EPB's AI smoke-recognition system catches a serial roadside waste burner in Dadu District", "summary": "fine under Article 32 of the Air Pollution Control Act for the open burning; CTS reports the fine as NT$100,000", "monetary_role": "reference_or_input_amount", "amount": { "source_text": "新臺幣10萬元以下 (up to NT$100,000)", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "TWD", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "reference_amount", "amount_basis": "reference", "period": "Detections July-August 2025; announcement September 13, 2025.", "unit": "one enforcement case", "beneficiary": "the 74-year-old burner pays", "payer": "Not established in the current source review.", "realization": "context", "accounting_boundary": "Legal maximum or reported fine, not a valuation of avoided pollution.", "comparator": "No monetary comparator.", "ai_attribution": "AI detected smoke; identification and enforcement depended on human investigation.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.epb.taichung.gov.tw/3092215/post", "title": "Taichung EPB's AI smoke-recognition system catches a serial roadside waste burner in Dadu District", "quote": "依違反空氣污染防制法第32條規定，處新臺幣10萬元以下罰鍰", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Resolve up-to versus actual NT$100,000 penalty", "Payment status", "Environmental outcome and any defensible valuation" ], "overlap_flags": [ "maximum_vs_imposed_conflict", "possible_penalty_not_outcome_amount", "transfer_perspective" ], "curation_status": "context_only", "source_review_status": "context_only", "case_link_status": "unresolved", "review_note": "The monetary endpoint is not clear enough to classify as government revenue on the existing fields.", "reviewed_on": null, "legacy_audit": { "monetary_role": "reference_or_input_amount", "money_type": "other", "money_subtype": "statutory maximum penalty", "claim_status": "possible_penalty_ceiling_with_conflicting_report", "currency": "TWD", "currency_basis": "explicit", "unit_basis": "one enforcement case", "time_basis": "Detections July-August 2025; announcement September 13, 2025.", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "No monetary comparator.", "ai_and_value_scope": "AI detected smoke; identification and enforcement depended on human investigation.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Legal maximum or reported fine, not a valuation of avoided pollution.", "net_benefit_established": false, "comparability_flags": [ "maximum_vs_imposed_conflict", "possible_penalty_not_outcome_amount", "transfer_perspective" ], "data_gaps": [ "Resolve up-to versus actual NT$100,000 penalty", "Payment status", "Environmental outcome and any defensible valuation" ], "extraction_note": "The monetary endpoint is not clear enough to classify as government revenue on the existing fields.", "source_review_required": true, "id": "AIBD-0147", "title": "Taichung EPB's AI smoke-recognition system catches a serial roadside waste burner in Dadu District", "value": { "amount": "新臺幣10萬元以下 (up to NT$100,000)", "what": "fine under Article 32 of the Air Pollution Control Act for the open burning; CTS reports the fine as NT$100,000", "who": "the 74-year-old burner pays", "estimate": true, "quote": "依違反空氣污染防制法第32條規定，處新臺幣10萬元以下罰鍰", "url": "https://www.epb.taichung.gov.tw/3092215/post" }, "sourceURL": "https://www.epb.taichung.gov.tw/3092215/post", "date_observed_as_stored": "2025-07-04", "existing_estimate_flag": true, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0028", "case_id": "AIBD-0238", "title": "DARPA's AIxCC final: autonomous systems find 54 of 63 synthetic vulnerabilities, patch 43, and surface 18 real bugs in open-source code", "summary": "average cost per competition task for the systems' bug reports and patches, which DARPA calls a fraction of the cost of traditional methods", "monetary_role": "reference_or_input_amount", "amount": { "source_text": "$152", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "USD", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "reference_amount", "amount_basis": "reference", "period": "AIxCC final competition in 2025", "unit": "per competition task", "beneficiary": "DARPA (cost of running the competitors' cyber reasoning systems per task)", "payer": "Not established in the current source review.", "realization": "context", "accounting_boundary": "A cost input; inclusions such as training, development and staff time are unknown.", "comparator": "Structured comparator concerns prior-round vulnerability detection; traditional monetary cost is only described qualitatively.", "ai_attribution": "Cost of running the cyber-reasoning systems, not a monetary benefit amount.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.darpa.mil/news/2025/aixcc-results", "title": "DARPA's AIxCC final: autonomous systems find 54 of 63 synthetic vulnerabilities, patch 43, and surface 18 real bugs in open-source code", "quote": "Competitor CRSs proved they can create valuable bug reports and patches for a fraction of the cost of traditional methods, with an average cost per competition task of about $152.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Traditional cost for comparable tasks", "Which cost components are included", "Task success and quality denominators", "Generalizability beyond competition" ], "overlap_flags": [ "cost_not_benefit", "per_task", "comparison_cost_missing", "synthetic_task_scope" ], "curation_status": "context_only", "source_review_status": "context_only", "case_link_status": "unresolved", "review_note": "Useful for an economics extension, but $152 itself is not a saving.", "reviewed_on": null, "legacy_audit": { "monetary_role": "reference_or_input_amount", "money_type": "other", "money_subtype": "system operating cost per task", "claim_status": "reported_average_cost", "currency": "USD", "currency_basis": "inferred_from_US_program", "unit_basis": "per competition task", "time_basis": "AIxCC final competition in 2025", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "Structured comparator concerns prior-round vulnerability detection; traditional monetary cost is only described qualitatively.", "ai_and_value_scope": "Cost of running the cyber-reasoning systems, not a monetary benefit amount.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "A cost input; inclusions such as training, development and staff time are unknown.", "net_benefit_established": false, "comparability_flags": [ "cost_not_benefit", "per_task", "comparison_cost_missing", "synthetic_task_scope" ], "data_gaps": [ "Traditional cost for comparable tasks", "Which cost components are included", "Task success and quality denominators", "Generalizability beyond competition" ], "extraction_note": "Useful for an economics extension, but $152 itself is not a saving.", "source_review_required": true, "id": "AIBD-0238", "title": "DARPA's AIxCC final: autonomous systems find 54 of 63 synthetic vulnerabilities, patch 43, and surface 18 real bugs in open-source code", "value": { "amount": "$152", "what": "average cost per competition task for the systems' bug reports and patches, which DARPA calls a fraction of the cost of traditional methods", "who": "DARPA (cost of running the competitors' cyber reasoning systems per task)", "estimate": false, "quote": "Competitor CRSs proved they can create valuable bug reports and patches for a fraction of the cost of traditional methods, with an average cost per competition task of about $152.", "url": "https://www.darpa.mil/news/2025/aixcc-results" }, "sourceURL": "https://www.darpa.mil/news/2025/aixcc-results", "date_observed_as_stored": "2025-08-08", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0029", "case_id": "AIBD-0207", "title": "Sheng Siong's facial-recognition CCTV, given a shoplifter's face after a 10 September 2025 stock check at its Punggol Central outlet, matches her to six earlier wine thefts and alerts the manager's phone when she returns at noon the next day; staff stop her with three unscanned bottles, police arrest her, she repays the chain and is jailed eight days on 12 May 2026", "summary": "restitution paid by the shoplifter for the stolen bottles that were not recovered (three bottles were recovered at the interception)", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "about S$470", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "SGD", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "avoided_loss", "amount_basis": "unresolved", "period": "Thefts September 2-11, 2025; restitution reported by May 12, 2026.", "unit": "one defendant and series of thefts", "beneficiary": "Sheng Siong", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross restitution; investigation and system costs not deducted.", "comparator": "Manual-review versus alert workflow; no monetary counterfactual.", "ai_attribution": "AI matched prior thefts and return; human interception, prosecution and restitution followed.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.channelnewsasia.com/singapore/shoplifter-caught-sheng-siong-facial-recognition-tech-jail-6115421", "title": "Sheng Siong's facial-recognition CCTV, given a shoplifter's face after a 10 September 2025 stock check at its Punggol Central outlet, matches her to six earlier wine thefts and alerts the manager's phone when she returns at noon the next day; staff stop her with three unscanned bottles, police arrest her, she repays the chain and is jailed eight days on 12 May 2026", "quote": "The retail manager called the police, and the three bottles of wine were recovered. Tan also made full restitution of about S$470 for the remaining stolen bottles.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Date and proof of restitution", "AI contribution relative to human detection", "Operating and enforcement costs" ], "overlap_flags": [ "restitution_transfer", "single_series_of_events", "different_recovery_dates" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "The S$470 excludes three recovered bottles; retain distinct cash and in-kind recoveries.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "restitution for stolen goods", "claim_status": "reported_restitution_paid", "currency": "SGD", "currency_basis": "explicit", "unit_basis": "one defendant and series of thefts", "time_basis": "Thefts September 2-11, 2025; restitution reported by May 12, 2026.", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "Manual-review versus alert workflow; no monetary counterfactual.", "ai_and_value_scope": "AI matched prior thefts and return; human interception, prosecution and restitution followed.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross restitution; investigation and system costs not deducted.", "net_benefit_established": false, "comparability_flags": [ "restitution_transfer", "single_series_of_events", "different_recovery_dates" ], "data_gaps": [ "Date and proof of restitution", "AI contribution relative to human detection", "Operating and enforcement costs" ], "extraction_note": "The S$470 excludes three recovered bottles; retain distinct cash and in-kind recoveries.", "source_review_required": true, "id": "AIBD-0207", "title": "Sheng Siong's facial-recognition CCTV, given a shoplifter's face after a 10 September 2025 stock check at its Punggol Central outlet, matches her to six earlier wine thefts and alerts the manager's phone when she returns at noon the next day; staff stop her with three unscanned bottles, police arrest her, she repays the chain and is jailed eight days on 12 May 2026", "value": { "amount": "about S$470", "what": "restitution paid by the shoplifter for the stolen bottles that were not recovered (three bottles were recovered at the interception)", "who": "Sheng Siong", "estimate": false, "quote": "The retail manager called the police, and the three bottles of wine were recovered. Tan also made full restitution of about S$470 for the remaining stolen bottles.", "url": "https://www.channelnewsasia.com/singapore/shoplifter-caught-sheng-siong-facial-recognition-tech-jail-6115421" }, "sourceURL": "https://www.channelnewsasia.com/singapore/shoplifter-caught-sheng-siong-facial-recognition-tech-jail-6115421", "date_observed_as_stored": "2025-09-11", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0030", "case_id": "AIBD-0150", "title": "Lynn White of Long Beach, California, who lost an eviction jury trial with a lawyer, may have overturned the eviction on appeal with ChatGPT and Perplexity, which she used to find procedural errors and draft it, avoiding roughly $55,000 in penalties and over $18,000 in back rent", "summary": "penalties avoided, plus more than $18,000 in overdue rent, by overturning the eviction on appeal", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "roughly $55,000", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "USD", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "unresolved", "period": "Reported October 8, 2025 after several months; ruling date unknown.", "unit": "one eviction appeal", "beneficiary": "Lynn White", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Personal gross liability reduction, not net legal benefit; costs and counterparty effects unstated.", "comparator": "Prior jury loss versus later appeal success; nonexperimental and not an equivalent legal comparison.", "ai_attribution": "AI assisted research and drafting; legal rules and personal advocacy also contributed.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.nbcnews.com/tech/innovation/ai-chatgpt-court-law-legal-lawyer-self-represent-pro-se-attorney-rcna230401", "title": "Lynn White of Long Beach, California, who lost an eviction jury trial with a lawyer, may have overturned the eviction on appeal with ChatGPT and Perplexity, which she used to find procedural errors and draft it, avoiding roughly $55,000 in penalties and over $18,000 in back rent", "quote": "Several months of litigation later, White managed to overturn her eviction notice and avoid roughly $55,000 in penalties and more than $18,000 in overdue rent.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Court ruling and actual legal effect", "Whether rent was waived or deferred", "Separate penalties and rent", "Case date and full litigation costs" ], "overlap_flags": [ "multiple_amount_components", "court_record_missing", "deferral_vs_extinguishment_unclear", "beneficiary_specific_transfer" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "The amount field covers roughly $55,000; what/quote separately describe over $18,000 rent. Keep both claims separate pending legal documentation.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "personal penalties and rent liability avoided", "claim_status": "reported_liability_avoided_with_legal_outcome_uncertain", "currency": "USD", "currency_basis": "inferred_from_US_context", "unit_basis": "one eviction appeal", "time_basis": "Reported October 8, 2025 after several months; ruling date unknown.", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "Prior jury loss versus later appeal success; nonexperimental and not an equivalent legal comparison.", "ai_and_value_scope": "AI assisted research and drafting; legal rules and personal advocacy also contributed.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Personal gross liability reduction, not net legal benefit; costs and counterparty effects unstated.", "net_benefit_established": false, "comparability_flags": [ "multiple_amount_components", "court_record_missing", "deferral_vs_extinguishment_unclear", "beneficiary_specific_transfer" ], "data_gaps": [ "Court ruling and actual legal effect", "Whether rent was waived or deferred", "Separate penalties and rent", "Case date and full litigation costs" ], "extraction_note": "The amount field covers roughly $55,000; what/quote separately describe over $18,000 rent. Keep both claims separate pending legal documentation.", "source_review_required": true, "id": "AIBD-0150", "title": "Lynn White of Long Beach, California, who lost an eviction jury trial with a lawyer, may have overturned the eviction on appeal with ChatGPT and Perplexity, which she used to find procedural errors and draft it, avoiding roughly $55,000 in penalties and over $18,000 in back rent", "value": { "amount": "roughly $55,000", "what": "penalties avoided, plus more than $18,000 in overdue rent, by overturning the eviction on appeal", "who": "Lynn White", "estimate": false, "quote": "Several months of litigation later, White managed to overturn her eviction notice and avoid roughly $55,000 in penalties and more than $18,000 in overdue rent.", "url": "https://www.nbcnews.com/tech/innovation/ai-chatgpt-court-law-legal-lawyer-self-represent-pro-se-attorney-rcna230401" }, "sourceURL": "https://www.nbcnews.com/tech/innovation/ai-chatgpt-court-law-legal-lawyer-self-represent-pro-se-attorney-rcna230401", "date_observed_as_stored": "2025-10-08", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0031", "case_id": "AIBD-0184", "title": "Estimated Universal Credit overpayments avoided", "summary": "The NAO reports DWP’s estimate that its advances model avoided £4.4 million of overpayments over three fiscal years.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "Estimated £4.4 million", "value": "4400000", "lower": null, "upper": null, "qualifier": "approximately", "currency": "GBP", "currency_basis": "context", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "avoided_loss", "amount_basis": "avoided_expenditure", "period": "U.K. fiscal years 2022–23 through 2024–25.", "unit": "Aggregate Universal Credit advances model estimate.", "beneficiary": "U.K. Department for Work and Pensions and public funds.", "payer": "DWP benefit expenditure that would otherwise have been overpaid.", "realization": "retrospective_estimate", "accounting_boundary": "Estimated gross overpayments avoided. DWP could not separate this model’s costs from wider analytics costs.", "comparator": "DWP’s estimate of payments that otherwise would have been made; method is not detailed in the cited passage.", "ai_attribution": "The model flagged higher-risk applications for staff assessment. The estimate concerns the combined screening and human-review process.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://www.nao.org.uk/wp-content/uploads/2025/10/Tackling-benefit-overpayments-due-to-fraud-and-error.pdf", "title": "Tackling benefit overpayments due to fraud and error", "quote": "the model generated savings of £4.4 million that would otherwise have been overpaid.", "date": "2025-10-22", "primary": true, "written_by": "independent", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "PDF page 31, paragraph 2.11" } ], "limitations": [ "An independent report relays DWP’s estimate; it does not establish a cost-adjusted AI return." ], "overlap_flags": [ "The model estimate must not be added to broader DWP totals that already include it." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "benefit overpayments avoided", "claim_status": "retrospective_estimate", "currency": "GBP", "currency_basis": "symbol_and_UK_context", "unit_basis": "DWP Universal Credit advance model", "time_basis": "2022-23 through 2024-25", "time_basis_completeness": "explicit_window", "monetary_comparator_assessment": "Randomized control sample supports risk-identification comparison; the monetary calculation is not set out in the existing fields.", "ai_and_value_scope": "Model prioritizes human caseworker review.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross overpayments avoided; account explicitly says model costs cannot be disaggregated.", "net_benefit_established": false, "comparability_flags": [ "gross_costs_not_separable", "operator_estimate_in_audit", "multi_year_cumulative", "distributional_effects" ], "data_gaps": [ "Saving-calculation method", "Treatment/control monetary outcomes", "Costs if separable", "False-positive and delay effects" ], "extraction_note": "Independent reporting of an agency estimate should not be mislabeled as an independently verified net saving.", "source_review_required": true, "id": "AIBD-0184", "title": "DWP's Universal Credit advances machine-learning model saved an estimated £4.4 million from 2022–23 to 2024–25", "value": { "amount": "£4.4 million", "what": "advances that would otherwise have been overpaid, saved from 2022-23 to 2024-25", "who": "Department for Work & Pensions", "estimate": true, "quote": "DWP estimates that, from 2022-23 to 2024-25, the model generated savings of £4.4 million that would otherwise have been overpaid.", "url": "https://www.nao.org.uk/wp-content/uploads/2025/10/Tackling-benefit-overpayments-due-to-fraud-and-error.pdf" }, "sourceURL": "https://www.nao.org.uk/wp-content/uploads/2025/10/Tackling-benefit-overpayments-due-to-fraud-and-error.pdf", "date_observed_as_stored": "2025-10-22", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0032", "case_id": "AIBD-0163", "title": "Quartz Solar annual imbalance-cost estimate", "summary": "Open Climate Fix estimates that Quartz Solar avoids at least £30 million in electricity imbalance costs per year.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "At least £30 million per year", "value": "30000000", "lower": null, "upper": null, "qualifier": "at_least", "currency": "GBP", "currency_basis": "explicit", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "cash_saving", "amount_basis": "avoided_expenditure", "period": "Annualized estimate published November 6, 2025; no completed measurement year is specified.", "unit": "Annual electricity imbalance costs.", "beneficiary": "Great Britain’s electricity system and its cost bearers.", "payer": "Electricity balancing and imbalance settlement participants.", "realization": "annualized_estimate", "accounting_boundary": "Estimated annual gross operating-cost avoidance, without implementation and ongoing service costs.", "comparator": "Previous NESO forecasting tools; the monetary calculation is not provided.", "ai_attribution": "Quartz Solar combines machine learning, satellite imagery, and weather data within NESO’s operating workflow. Its developer supplies the monetary estimate.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://www.openclimatefix.org/insights/neso-adopts-ai-solar-forecasting-control-room", "title": "National Energy System Operator adopts AI solar forecasting", "quote": "helping to avoid at least £30 million in imbalance costs each year", "date": "2025-11-06", "primary": true, "written_by": "makers", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": null } ], "limitations": [ "An annualized developer estimate is not a verified cash saving for a completed year." ], "overlap_flags": [ "Future scenarios or cumulative savings cannot be added to this annual rate without a defined period." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "cash_saving", "money_subtype": "electricity imbalance costs", "claim_status": "annualized_developer_estimate", "currency": "GBP", "currency_basis": "explicit", "unit_basis": "per year at current adoption", "time_basis": "Annualized estimate announced November 2025; measured savings interval unspecified.", "time_basis_completeness": "annualized_without_window", "monetary_comparator_assessment": "Forecast errors reportedly halved; no operator monetary baseline.", "ai_and_value_scope": "Maker estimates value of an integrated forecast service; operator confirms deployment, not the saving figure.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross avoided imbalance cost; service and integration costs unstated.", "net_benefit_established": false, "comparability_flags": [ "annualized_not_observed_year", "developer_estimate", "operational_to_financial_conversion", "future_scenario_separate" ], "data_gaps": [ "Calculation from forecast error to cost", "Operator confirmation", "Observed period", "Costs and consumer pass-through" ], "extraction_note": "Keep the £30m annual estimate separate from the £150m/year conditional 2035 scenario.", "source_review_required": true, "id": "AIBD-0163", "title": "Open Climate Fix's Quartz Solar goes live in NESO's control room and may avoid £30m a year in imbalance costs", "value": { "amount": "GBP 30 million", "what": "imbalance cost avoided per year", "who": "Great Britain's electricity system operator NESO and, through balancing costs, consumers", "estimate": true, "quote": "Since adoption, Quartz Solar has halved forecasting errors, according to Open Climate Fix, equivalent to saving at least GBP 30 million in imbalance cost per year, with potential to increase to GBP 150 million ($197.0 million) per year in 2035 if government solar capacity targets are met.", "url": "https://www.pv-magazine.com/2025/11/07/ai-powered-solar-forecasting-helps-uk-grid-operator-reduce-balancing-costs/" }, "sourceURL": "https://www.pv-magazine.com/2025/11/07/ai-powered-solar-forecasting-helps-uk-grid-operator-reduce-balancing-costs/", "date_observed_as_stored": "2025-11-06", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0033", "case_id": "AIBD-0181", "title": "Ghana Revenue Authority launches Publican AI trade analytics, which may have recovered GH¢15 million", "summary": "recovered from fraudulent importers", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "GH¢15 million", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "GHS", "currency_basis": "explicit", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "revenue", "amount_basis": "unresolved", "period": "Reported January 2026; recovery interval unspecified.", "unit": "program recovery to date", "beneficiary": "Government of Ghana (Ghana Revenue Authority)", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Gross recovery claim; costs and refund/reversal adjustments are absent.", "comparator": "Assessment-time comparison only; claimed revenue-gain percentages refer to pilots.", "ai_attribution": "AI analytics claim and recovery appear in a registry summary; source account does not establish mechanism or incremental share.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://oecd.ai/en/incidents/2026-01-16-e77d", "title": "Ghana Revenue Authority launches Publican AI trade analytics, which may have recovered GH¢15 million", "quote": "The system has already enabled the government to recover GH¢15 million from fraudulent importers and aims to address significant revenue leakages in Ghana's customs operations.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Original GhanaWeb source and authority confirmation", "Recovery dates and accounting definition", "AI role", "Costs" ], "overlap_flags": [ "source_chain_needs_review", "recovery_period_missing", "pilot_vs_rollout_timing" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "The 40-45% pilot revenue claim and the GH¢15m recovered need separate claim and cohort identifiers.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "revenue", "money_subtype": "customs fraud recovery", "claim_status": "reported_recovery_via_secondary_summary", "currency": "GHS", "currency_basis": "explicit", "unit_basis": "program recovery to date", "time_basis": "Reported January 2026; recovery interval unspecified.", "time_basis_completeness": "incomplete_window", "monetary_comparator_assessment": "Assessment-time comparison only; claimed revenue-gain percentages refer to pilots.", "ai_and_value_scope": "AI analytics claim and recovery appear in a registry summary; source account does not establish mechanism or incremental share.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross recovery claim; costs and refund/reversal adjustments are absent.", "net_benefit_established": false, "comparability_flags": [ "source_chain_needs_review", "recovery_period_missing", "pilot_vs_rollout_timing" ], "data_gaps": [ "Original GhanaWeb source and authority confirmation", "Recovery dates and accounting definition", "AI role", "Costs" ], "extraction_note": "The 40-45% pilot revenue claim and the GH¢15m recovered need separate claim and cohort identifiers.", "source_review_required": true, "id": "AIBD-0181", "title": "Ghana Revenue Authority launches Publican AI trade analytics, which may have recovered GH¢15 million", "value": { "amount": "GH¢15 million", "what": "recovered from fraudulent importers", "who": "Government of Ghana (Ghana Revenue Authority)", "estimate": false, "quote": "The system has already enabled the government to recover GH¢15 million from fraudulent importers and aims to address significant revenue leakages in Ghana's customs operations.", "url": "https://oecd.ai/en/incidents/2026-01-16-e77d" }, "sourceURL": "https://oecd.ai/en/incidents/2026-01-16-e77d", "date_observed_as_stored": "2026-01-16", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "high", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0034", "case_id": "AIBD-0202", "title": "Gyeonggi ARES AI pig-weighing pilot cuts shipping age 5.6 days and lifts income per head 18% on six farms", "summary": "income per pig shipped with the AI shipping selection system, up from 10만 원 (100,000 won) per head, an 18% increase", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "KRW118,000 per pig (reported level)", "value": "118000", "lower": null, "upper": null, "qualifier": "exact", "currency": "KRW", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "other", "amount_basis": "outcome_level", "period": "2025 pilot, reported March 2026; within-year observation length unspecified.", "unit": "per pig shipped across six pilot farms", "beneficiary": "the six pilot pig farms in Anseong, Pocheon and Yeoju", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "The source term income does not establish revenue, gross margin or net profit.", "comparator": "Explicit 100,000 won before versus 118,000 won after, an 18% increase.", "ai_attribution": "Uncontrolled before/after pilot combines weighing, shipping advice and farm responses.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://news.tf.co.kr/read/national/2299220.htm", "title": "Gyeonggi ARES AI pig-weighing pilot cuts shipping age 5.6 days and lifts income per head 18% on six farms", "quote": "출하일령이 182.3일에서 176.7일로 5.6일 줄었다. 또 1+등급 출하율은 35.3%에서 37.7%으로 2.4%p 증가하고, 두당 소득이 10만 원에서 11만 8000원으로 18% 늘었다.", "date": null, "primary": false, "written_by": "unknown", "reviewed_at": null, "review_method": "unreviewed", "locator": null } ], "limitations": [ "Definition of income and included costs", "Farm/sample weighting and observation period", "Other changes and market prices", "System cost" ], "overlap_flags": [ "outcome_level_not_increment", "income_definition_missing", "per_animal", "uncontrolled_before_after" ], "curation_status": "awaiting_review", "source_review_status": "awaiting_review", "case_link_status": "unresolved", "review_note": "The KRW118,000 figure is a post-intervention per-pig income level, not the increment. KRW100,000 is the reported earlier level; any derived difference needs a verified common basis and attribution review.", "reviewed_on": null, "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "other", "money_subtype": "farm income per pig; revenue versus profit unresolved", "claim_status": "reported_before_after_income_level", "currency": "KRW", "currency_basis": "currency_inferred_from_Korean_context", "unit_basis": "per pig shipped across six pilot farms", "time_basis": "2025 pilot, reported March 2026; within-year observation length unspecified.", "time_basis_completeness": "partial_window", "monetary_comparator_assessment": "Explicit 100,000 won before versus 118,000 won after, an 18% increase.", "ai_and_value_scope": "Uncontrolled before/after pilot combines weighing, shipping advice and farm responses.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "The source term income does not establish revenue, gross margin or net profit.", "net_benefit_established": false, "comparability_flags": [ "outcome_level_not_increment", "income_definition_missing", "per_animal", "uncontrolled_before_after" ], "data_gaps": [ "Definition of income and included costs", "Farm/sample weighting and observation period", "Other changes and market prices", "System cost" ], "extraction_note": "118,000 won is the post-intervention level. The implied 18,000 won difference is an arithmetic derivation, not another source amount, and should be labeled if used.", "source_review_required": true, "id": "AIBD-0202", "title": "Gyeonggi ARES AI pig-weighing pilot cuts shipping age 5.6 days and lifts income per head 18% on six farms", "value": { "amount": "11만 8000원", "what": "income per pig shipped with the AI shipping selection system, up from 10만 원 (100,000 won) per head, an 18% increase", "who": "the six pilot pig farms in Anseong, Pocheon and Yeoju", "estimate": false, "quote": "출하일령이 182.3일에서 176.7일로 5.6일 줄었다. 또 1+등급 출하율은 35.3%에서 37.7%으로 2.4%p 증가하고, 두당 소득이 10만 원에서 11만 8000원으로 18% 늘었다.", "url": "https://news.tf.co.kr/read/national/2299220.htm" }, "sourceURL": "https://news.tf.co.kr/read/national/2299220.htm", "date_observed_as_stored": "2026-03-06", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0035", "case_id": "AIBD-0210", "title": "Retirement holdings in TIAA’s reported scam intervention", "summary": "Reported holdings; the attempted withdrawal amount is unspecified.", "monetary_role": "outcome_amount_claim", "amount": { "source_text": "over $3 million", "value": null, "lower": null, "upper": null, "qualifier": "unparsed", "currency": "USD", "currency_basis": "context", "price_basis": "Price year and inflation basis await review." }, "additional_amounts": [], "financial_type": "avoided_loss", "amount_basis": "unresolved", "period": "Recently before June 17, 2026; exact date unstated.", "unit": "one participant’s retirement holdings", "beneficiary": "the 76-year-old TIAA participant", "payer": "Not established in the current source review.", "realization": "unverified", "accounting_boundary": "Reported holdings; the attempted withdrawal amount is unspecified.", "comparator": "No counterfactual; the withdrawal could also have triggered ordinary review.", "ai_attribution": "AI flag started a human intervention chain; the CEO states that AI alone would not necessarily have protected the participant.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://fortune.com/2026/06/22/tiaa-ceo-thasunda-brown-duckett-retirement-crisis-annuity-income-america/", "title": "Interview with Thasunda Brown Duckett", "quote": "had over $3 million", "date": "2026-06-22", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-24", "review_method": "full_text", "locator": "“How TIAA is approaching AI—and whether it threatens the retirement market,” participant example." } ], "limitations": [ "Independent case confirmation", "Event date, attempted withdrawal amount, and transfer records", "Counterfactual risk and recovery potential", "Intervention cost" ], "overlap_flags": [ "holdings_not_verified_avoided_loss", "single_unnamed_case", "operator_anecdote", "human_intervention_chain" ], "curation_status": "awaiting_review", "source_review_status": "primary_source_checked", "case_link_status": "unresolved", "review_note": "Withdrawal amount unverified.", "reviewed_on": "2026-10-24", "legacy_audit": { "monetary_role": "outcome_amount_claim", "money_type": "avoided_loss", "money_subtype": "attempted scam withdrawal stopped", "claim_status": "reported_prevention_of_nominal_attempt", "currency": "USD", "currency_basis": "inferred_from_US_context", "unit_basis": "one participant withdrawal request", "time_basis": "Recently before June 17, 2026; exact date unstated.", "time_basis_completeness": "undated_one_off", "monetary_comparator_assessment": "No counterfactual; the withdrawal could also have triggered ordinary review.", "ai_and_value_scope": "AI flag started a human intervention chain; the CEO states that AI alone would not necessarily have protected the participant.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Nominal portfolio at risk; expected loss and intervention costs unknown.", "net_benefit_established": false, "comparability_flags": [ "attempted_amount_not_expected_loss", "single_unnamed_case", "operator_anecdote", "human_intervention_chain" ], "data_gaps": [ "Independent case confirmation", "Event date and transfer status", "Counterfactual risk and recovery potential", "Intervention cost" ], "extraction_note": "The entire portfolio is the exposure blocked, not a statistically estimated expected saving.", "source_review_required": true, "id": "AIBD-0210", "title": "TIAA's AI may have flagged a 76-year-old participant's out-of-pattern request to withdraw his entire $3 million retirement portfolio; the fraud team and his daughter persuaded him it was a scam and the transfer was stopped", "value": { "amount": "$3 million", "what": "retirement portfolio whose withdrawal to a scam was stopped", "who": "the 76-year-old TIAA participant", "estimate": false, "quote": "Recently, TIAA flagged an unusual request from a 76-year-old customer attempting to withdraw his entire $3 million retirement portfolio.", "url": "https://fortune.com/2026/06/17/tiaa-saved-retiree-from-scam-using-artifical-intelligence-human-centered-work-ceo-thasunda-brown-duckett/" }, "sourceURL": "https://fortune.com/2026/06/17/tiaa-saved-retiree-from-scam-using-artifical-intelligence-human-centered-work-ceo-thasunda-brown-duckett/", "date_observed_as_stored": "2026-06-17", "existing_estimate_flag": false, "existing_comparator_present": false, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false }, { "id": "DIV-0036", "case_id": "AIBD-0262", "title": "Payment for Nimbus’s TYK2 program", "summary": "Takeda’s annual securities report states that it paid Nimbus $4 billion upfront for the subsidiary owning the TYK2 program.", "monetary_role": "commercial_transaction", "amount": { "source_text": "US$4 billion", "value": "4000000000", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "explicit", "price_basis": "Nominal source amount; price year not separately stated." }, "additional_amounts": [], "financial_type": "commercial_receipt", "amount_basis": "transaction", "period": "Acquisition closed February 8, 2023; payment is reported in the fiscal-2023 annual securities report filed June 26, 2024.", "unit": "One acquisition’s upfront consideration.", "beneficiary": "Nimbus Therapeutics and the relevant sellers.", "payer": "Takeda Pharmaceutical Company.", "realization": "paid", "accounting_boundary": "Gross acquisition consideration, not drug revenue, seller profit, clinical benefit, or newly created societal wealth.", "comparator": "No estimate of the transaction value without AI-assisted design is supplied.", "ai_attribution": "The acquisition concerns the NDI-034858/TAK-279 program linked to the AIBD computational-design case. The transaction source does not value AI’s incremental contribution.", "ai_attribution_status": "ai_component_supported", "net_benefit_established": false, "sources": [ { "url": "https://assets-dam.takeda.com/image/upload/Global/Investor/ASR/Y2023/E_FY23_ASR.pdf", "title": "Takeda annual securities report for the year ended March 31, 2024", "quote": "we paid Nimbus USD 4.0 billion upfront following the closing of the transaction", "date": "2024-06-26", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-23", "review_method": "full_text", "locator": "Printed page 50, material contracts" } ], "limitations": [ "The 2023 transaction predates the later clinical milestones in the parent AIBD record.", "No AI-specific premium or net economic benefit is established." ], "overlap_flags": [ "Two contingent $1 billion milestones are excluded; the $4 billion payment is not the $6 billion maximum deal value." ], "curation_status": "curated", "source_review_status": "primary_source_checked", "case_link_status": "matched", "review_note": "The cited monetary statement and its accounting boundary were checked. This is not a certification of net benefit or AI-specific causation.", "reviewed_on": "2026-10-23", "legacy_audit": { "monetary_role": "commercial_transaction", "money_type": "revenue", "money_subtype": "upfront sale of drug rights", "claim_status": "announced_transaction_payment", "currency": "USD", "currency_basis": "inferred_from_US_biopharma_reporting", "unit_basis": "one rights acquisition", "time_basis": "Deal announced December 13, 2022; main clinical outcome occurs in 2026.", "time_basis_completeness": "dated_one_off", "monetary_comparator_assessment": "Clinical comparator is efficacy, not value or development cost.", "ai_and_value_scope": "Sale of a compound developed with computational assistance; the AI-attributable increment in commercial value is unknown.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Gross transaction consideration, not profit, patient benefit, or net social value.", "net_benefit_established": false, "comparability_flags": [ "commercial_transfer", "transaction_vs_clinical_outcome", "payment_completion_not_checked", "AI_share_unresolved" ], "data_gaps": [ "Deal closing and actual payment", "Development and acquisition costs", "AI contribution to transaction value", "Separate optional milestones from upfront payment" ], "extraction_note": "The $4bn may belong in a commercial-outcomes view, but it is not a monetary measure of the clinical benefit.", "source_review_required": true, "id": "AIBD-0262", "title": "Zasocitinib, computationally designed by Nimbus with Schrodinger's platform, beats deucravacitinib in phase 3; FDA priority review", "value": { "amount": "$4 billion", "what": "upfront payment for rights to the computationally designed compound", "who": "Nimbus (seller), paid by Takeda", "estimate": false, "quote": "The Japanese pharmaceutical company will pay Nimbus $4 billion upfront for rights to the medicine, which is in clinical testing for multiple autoimmune conditions.", "url": "https://www.biopharmadive.com/news/nimbus-takeda-tyk2-inhibitor-drug-deal/638614/" }, "sourceURL": "https://www.biopharmadive.com/news/nimbus-takeda-tyk2-inhibitor-drug-deal/638614/", "date_observed_as_stored": "2026-09-30", "existing_estimate_flag": false, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": true, "original_quote_and_source_link_present": true }, "eligible": true }, { "id": "DIV-0037", "case_id": "AIBD-0359", "title": "VTTI/VDOT thermal-camera AI wildlife detection pilot on Route 460 near Grundy, Virginia: 97% detection accuracy and a 3 mph average speed drop at activated signs", "summary": "Estimated comprehensive societal cost per elk crash, including economic losses and monetized pain and suffering. No AI-attributable avoided crash or saving is established.", "monetary_role": "reference_or_input_amount", "amount": { "source_text": "$80,771.49 per elk crash", "value": "80771.49", "lower": null, "upper": null, "qualifier": "exact", "currency": "USD", "currency_basis": "context", "price_basis": "2020 U.S. dollars. The cited VDOT memorandum labels its unit costs as 2020 values and explains that it uses 2020 economic indicators. The weighted calculation reproduces the reported estimate." }, "additional_amounts": [], "financial_type": "reference_amount", "amount_basis": "reference", "period": "Crash records from five states cover different periods between 2009 and 2021. The estimate was published in September 2023 and uses 2020 U.S. dollar valuations.", "unit": "per elk collision", "beneficiary": "Road users and society bear the economic and human costs represented by this valuation.", "payer": "Not established in the current source review.", "realization": "context", "accounting_boundary": "Comprehensive societal valuation per elk crash. Includes tangible economic impacts and monetized pain and suffering. This is neither a cash payment nor a measured benefit of AI.", "comparator": "Measured speed reduction, but the record explicitly states that prevented crashes were not established.", "ai_attribution": "AI triggers alerts; a crash-reduction or loss-avoidance quantity is not supplied.", "ai_attribution_status": "ai_component_unresolved", "net_benefit_established": false, "sources": [ { "url": "https://www.vdot.virginia.gov/media/vdotvirginiagov/projects/bristol/corridor-q-driver-safety-and-maintaining-habitat/VTRC-Evaluation-of-Wildlife-Crossing-Design-for-Corridor-Q-Final-Report_acc09032025_PM.pdf", "title": "Donaldson et al. An Evaluation of Wildlife Crossing Design, Placement, Costs, and Funding Opportunities for Corridor Q. VTRC 24-R8, September 2023.", "quote": "Average elk crash cost", "date": null, "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-24", "review_method": "full_text", "locator": "pp. 41–42, table 12 and note c; the calculated value is $80,771.49." }, { "url": "https://loudouncoalition.org/wp-content/uploads/2022/06/37-VDOT-Crash-Costs-Memo_acc050222-1.pdf", "title": "Cole, Mark A. Memorandum: VDOT Traffic Crash Costs—Version 1. Virginia Department of Transportation, March 2, 2022.", "quote": "Virginia KABCO Crash Unit Costs (2020)", "date": "2022-03-02", "primary": true, "written_by": "deployer", "reviewed_at": "2026-10-24", "review_method": "full_text", "locator": "1, tables 1–2; 3, table 6 and price adjustment method. Preserved copy of the original VDOT memorandum." } ], "limitations": [ "No quantity of collisions prevented by AI", "Severity mix drawn from reported crashes in five other states", "Monetized pain and suffering is not a cash expense", "System costs and any net AI benefit not established", "For 104 of 171 California crash records, injury severity was missing and was treated as property damage only in the calculation." ], "overlap_flags": [ "harm_unit_cost_not_avoided_loss", "missing_avoided_quantity", "estimate_not_benefit_projection" ], "curation_status": "context_only", "source_review_status": "context_only", "case_link_status": "unresolved", "review_note": "Traced the Virginia Tech article through the 2024 feasibility study to the original 2023 analysis and the cited 2022 VDOT cost memorandum. Table 12 uses 3,421 reported elk collisions and Virginia severity valuations. The memorandum establishes 2020 prices, and its costs reproduce $80,771.49 after rounding. No benefit of the AI pilot is estimated.", "reviewed_on": "2026-10-24", "legacy_audit": { "monetary_role": "reference_or_input_amount", "money_type": "other", "money_subtype": "reference cost of an elk collision", "claim_status": "modeled_unit_cost_without_avoided_event_count", "currency": "USD", "currency_basis": "inferred_from_US_context", "unit_basis": "per elk collision", "time_basis": "Feasibility-study cost; valuation period and price year absent.", "time_basis_completeness": "reference_period_missing", "monetary_comparator_assessment": "Measured speed reduction, but the record explicitly states that prevented crashes were not established.", "ai_and_value_scope": "AI triggers alerts; a crash-reduction or loss-avoidance quantity is not supplied.", "gross_net_status": "not_explicitly_established", "gross_net_interpretation": "Cost of a harm event, not a benefit amount; inclusions in that cost estimate unknown.", "net_benefit_established": false, "comparability_flags": [ "harm_unit_cost_not_avoided_loss", "missing_avoided_quantity", "estimate_not_benefit_projection" ], "data_gaps": [ "Expected or observed collisions prevented", "Cost components and price year", "Forecast horizon and uncertainty", "System costs" ], "extraction_note": "Do not multiply the crash cost by an invented prevented-crash count. A genuine projected benefit would need an explicit frequency/risk model.", "source_review_required": true, "id": "AIBD-0359", "title": "VTTI/VDOT thermal-camera AI wildlife detection pilot on Route 460 near Grundy, Virginia: 97% detection accuracy and a 3 mph average speed drop at activated signs", "value": { "amount": "more than $80,000 per crash", "what": "average cost of one elk collision, the harm the system aims to avert; the study did not establish that any collision was prevented", "who": "motorists and the state bear the cost per crash", "estimate": true, "quote": "A feasibility study associated with the project estimated that elk collisions cost more than $80,000 per crash on average.", "url": "https://bioengineer.org/virginia-pilot-links-wildlife-alerts-to-3-mph-speed-reduction/" }, "sourceURL": "https://bioengineer.org/virginia-pilot-links-wildlife-alerts-to-3-mph-speed-reduction/", "date_observed_as_stored": "2026-10-20", "existing_estimate_flag": true, "existing_comparator_present": true, "source_verification_status": "not_reopened_or_verified_in_this_audit", "reviewed_existing_fields": [ "value", "title", "metric", "comparator", "account", "ai_role", "date_observed" ], "source_review_priority": "standard", "unit_basis_present": true, "amount_specific_measurement_window_fully_explicit": false, "original_quote_and_source_link_present": true }, "eligible": false } ]