What happened
In August 2023 Australia's Department of Health, Disability and Ageing began building a system to detect fraud and serious non-compliance in Medicare claims. One of its six models, which the department counted as AI-enabled, ran from July 2024 to December 2025. The Australian National Audit Office (ANAO) examined it in an audit published on 21 September 2026. The ANAO found that eight potential fraud or non-compliance matters came in part from the AI-enabled model. By February 2026 five had been confirmed as potential fraud or non-compliance and referred for investigation or audit, and no money had been recovered. In a preliminary analysis of August 2025, the department found that the model identified matters accurately 22% of the time. About 80% of the providers that it flagged were new to the other models. The ANAO notes that the eight matters, estimated at $5,253,601, combined results from several models, including models without AI. The department stopped using the AI-enabled model after December 2025 because manual assessment could not keep up with its volume of matters.
Sources
-
Audit report, by authors with no tie to the system.
Australian National Audit Office. Artificial Intelligence and Medicare Benefits Integrity. Auditor-General Report No. 5 of 2026-27, published 21 September 2026. Paragraph 19 and paragraphs 5.5 to 5.14 (archived copy of the report page).
Eight potential non-compliance or fraud matters were created and referred for preliminary analysis using, in part, the AI-enabled model in the MBS fraud detection system. While ad hoc preliminary performance analysis showed the AI-enabled model was efficacious, it was not used after December 2025
Details
AI system
- Workflow
-
The model analyzed Medicare provider claiming data for behaviors known to be linked to fraud. Department officials manually assessed the model's signals and referred matters of concern for further analysis and investigation.
Money
- Kind of value
-
Estimated value of potential fraud or non-compliance in the eight matters identified in part with the AI-enabled model, whose results also drew on models without AI
- Amount
-
Estimate: $5,253,601
- Recipients and payers
-
Medicare, the Australian Government's health benefits program
Follow-up
- Conditions for replication
-
Staff must be able to assess manually the volume of matters that the model generates. Referred matters must be confirmed through analysis, investigation or audit before any money is recovered.
Entry details
- Sector
-
Health insurance fraud and improper payments.
- Purpose
-
The Department of Health, Disability and Ageing built the AI-enabled model to focus manual reviews on the riskier of the many providers whose Medicare claiming showed behaviors linked to fraud.
38 more questions not yet checked
model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, currency and prices, period and unit, realization, gross or net, included costs, comparison, method, return and payback, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Medicare's AI-enabled model flags eight possible fraud cases (AIBD-1009). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-1009.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)