AI system
- System
-
ALICE (Analisador de Licitações, Contratos e Editais), CGU text-mining and AI alert tool
- Model
-
Not assessed
- Release
-
Not assessed
- Application developer
-
Controladoria-Geral da União (CGU), in-house - Role
-
Not assessed
- Configuration
-
Not assessed
- Inputs and outputs
-
Not assessed
- Autonomy
-
Not assessed
- Workflow
-
ALICE audit trails that flag price surveys with abnormally high estimates in documents published on Compras.gov.br and bank procurement systems. CGU preventive audit teams that examine flagged tenders before the auction concludes and recommend corrections. Managers' response to audit recommendations (suspension, revocation or new price estimate).
- Human oversight
-
Not assessed
- Relation to AI
-
The CGU release states that each tender “teve alerta do Robô ALICE” and that auditors then confirmed the overpricing risk. The suspension, revocation and re-estimation are described as results of that audit action. Human auditors and hospital managers made the decisions. ALICE's role was detection and triage. Developer and deployer are the same public body, so this is the body documenting its own tool. The CGU page is on record as an archived copy, because the live page was restricted during Brazil's 2026 electoral period.
- AI contribution
-
The CGU release states that each tender “teve alerta do Robô ALICE” and that auditors then confirmed the overpricing risk. The suspension, revocation and re-estimation are described as results of that audit action. Human auditors and hospital managers made the decisions. ALICE's role was detection and triage. Developer and deployer are the same public body, so this is the body documenting its own tool. The CGU page is on record as an archived copy, because the live page was restricted during Brazil's 2026 electoral period.
Benefits
- Outcome identifiers
-
Not assessed
- Benefit
-
a saving of R$3,063,774.84
- Category
-
Reduced Risk
- Status
-
Not assessed
- Beneficiary groups
-
People. Organizations. Society.
- Beneficiaries
-
Not assessed
- Reach
-
Not assessed
- Distribution and access
-
Not assessed
- Duration
-
Not assessed
- Expected?
-
Yes
- Costs and harms
-
Not assessed
Measurement
- Studies
-
Not assessed
- Study design
-
Not assessed
- Measurement
-
Not assessed
- Setting and sample
-
Not assessed
- Measures
-
a saving of R$3,063,774.84
- Comparison
-
the hospital's original price estimate for tender 90017/2024
- Results
-
Reference: valor estimado em R$ 15.450.714,80. Result: R$ 12.386.939,96.
- Additional value from AI
-
Not assessed
- Limitations
-
Not assessed
- Denominators
-
Not assessed
- Materials for replication
-
Not assessed
Money
- Kind of value
-
reduction of the estimated value of tender 90017/2024 after the re-estimate (the release gives no total for the three tenders)
- Amount
-
R$ 3.063.774,84
- Currency and prices
-
Not assessed
- Period and unit
-
Not assessed
- Realization
-
Not assessed
- Recipients and payers
-
Hospital Federal de Ipanema, federal budget
- Gross or net
-
Not assessed
- Included costs
-
Not assessed
- Comparison
-
Not assessed
- Method
-
Not assessed
- Return and payback
-
Not assessed
People and places
- Participants
-
Developer:
Controladoria-Geral da União (CGU), in-house. Operator: Controladoria-Geral da União (CGU). - Jurisdictions
-
Benefit: Rio de Janeiro (state),
Brazil. Developer: Brazil. Operator: Brazil. - Ownership and control
-
Not assessed
- Dependencies
-
Not assessed
- Cross-border connections
-
Not assessed
- Geopolitical implications
-
Not assessed
Follow-up
- Conditions for replication
-
Procurement planning documents and notices must be machine-readable in the monitored systems. Audit capacity must exist to act on alerts before award. Reference prices must be available to confirm overpricing.
- Replication attempts
-
3 follow-up entries. Benefit recurred: 3.
- Later findings
-
The two later press items (21 and 28 January 2025) restate the release; no later development was found,.
- Availability
-
Not assessed
- Institutional decision
-
Not assessed
- Grounds
-
Not assessed
- Action and result
-
Not assessed
- Success criterion
-
Not assessed
Evidence
- Source numbers
-
AIBD-0081-source-0001–0006
- Citations
-
Numbered references below.
- Language
-
Not assessed
- Source relationships
-
Developer, Independent
- Original evidence
-
Not assessed
- Access and versions
-
Not assessed
- Supported findings
-
Not assessed
- Evidence links
-
Not assessed
- Evidence strength
-
Not assessed
- Missing evidence
-
Not assessed
Grades
- Assessed benefits
-
People. Organizations. Society.
- Depth
-
People: Not assessed. Organizations: 2. Society: 1.
- Reasons
-
People: No identifiable individual beneficiary is named; the beneficiaries are public bodies and, indirectly, taxpayers. Organizations: Material to the two hospitals and the federal budget. One tender saw a documented R$3,063,774.84 reduction. One suspension and one revocation removed R$13.4 million of confirmed overpricing risk. Society: Small at national scale; three tenders among 118,069 processes analyzed in 2024.
- Observed or intended
-
Not assessed
- Rubric
-
Not assessed
- Assessor and date
-
Not assessed
- Alternative assessments
-
Not assessed
- Fill
-
Independently reported
Entry details
- Kind
-
Not assessed
- Related entries
-
Government fraud-risk flagging algorithm, like ALICE's tender alerts, sent 200,000 housing-benefit claims to investigation, two-thirds legitimate, wasting public funds and stressing claimants.
- Field
-
Public administration: government audit and procurement control. Federal internal control (anti-corruption and waste prevention).
- Purpose
-
Continuous automated audit of federal procurement notices to detect risks such as inflated price estimates before contracts are awarded
- Dates
-
Date basis unassessed: 2025 (precision unassessed). Updated: 21 October 2026.
- Stage
-
Not assessed
- Version
-
SHA-256: d6b958633d54…
References
- Other identifiers
-
AI Incident Database: 738.
- Category mappings
-
Not assessed
- Brief
-
Not assessed
- Revisions
-
9 revisions. Latest: 21 October 2026.
Sources
-
Report Developer
Oliveira, T. C., Rocha, A. L. M. & Rezende, M. S. (2022). Alice: Desafios, resultados e perspectivas da ferramenta de auditoria contínua de compras públicas governamentais com uso de inteligência artificial. Revista da CGU 14(26), jul-dez/2022, published 20 Dec 2022; a practitioner report (relato) in the CGU's own journal. doi:10.36428/revistadacgu.v14i26.530
Já são mais de R$ 9,7 bilhões em compras suspensas a partir de suas indicações, apenas pela CGU
-
News report Independent author
Grossmann, L. O. (2025). Robô da CGU rastreia R$ 25 bilhões em licitações. Convergência Digital, 15 Jan 2025.
Das auditorias concluídas, foram economizados R$ 257,25 milhões
-
News report Independent author
R7 Planalto (2025). ALICE: robô da CGU previne erros e desperdícios de licitações públicas, 28 Jan 2025, republished by Licitação.net.
Nos primeiros relatórios de auditoria de 2024, ALICE identificou riscos de sobrepreço em três licitações de Hospitais Federais no Rio de Janeiro, relacionadas à compra de equipamentos e insumos.
-
News report Independent author
Observatório da Nova Lei de Licitações (2025). Atuação do robô ALICE auxilia CGU a rastrear R$ 25 bilhões em licitações, 21 Jan 2025 (with information from CGU).
uma licitação para a compra de aventais laminados, inicialmente estimada em R$ 16,7 milhões, foi suspensa após a detecção de um possível sobrepreço de R$ 10,9 milhões
-
Press release Developer
Controladoria-Geral da União (2025). Atuação do Robô ALICE segue auxiliando auditores a realizar trabalhos preventivos, que ajudam a evitar desperdícios e irregularidades. gov.br/cgu, 14 Jan 2025.
a licitação 90016/2024 - Relatório 1686291 -– teve alerta do Robô ALICE e foi analisada pelos auditores que indicaram risco de sobrepreço de R$ 10.925.029,92
Como resultado da atuação, a licitação foi revogada.