IRS's Rapid AI Scale-Up for Audit Selection — GAO Finds a Twelvefold Jump in Use Cases and Serious Governance Gaps
United States of America
The IRS grew from 10 to 126 active AI use cases between 2022 and mid-2025 to flag high-risk tax returns …
Israel · Jerusalem · See the Israel profile · See the Jerusalem profile
Top 25% 73/100 · Ask Evidence Copilot about this practice
Coordinated by Israel's State Comptroller, 12 EUROSAI audit institutions applied a shared 9-topic, 92-question framework to benchmark government AI readiness, finding ownership ahead of operational guidance and ethics still principle-level, not tested assurance.
In March 2026, the Office of the State Comptroller and Ombudsman of Israel published the final consolidated report of the EUROSAI Parallel Audit on Artificial Intelligence, a coordinated exercise in which 12 Supreme Audit Institutions (SAIs) — Albania, Estonia, France, Israel, Italy, Latvia, Lithuania, North Macedonia, Poland, Romania, Slovakia and Switzerland — examined their national governments' preparedness for AI. SAI Israel led the process throughout, reflecting Matanyahu Englman's role as EUROSAI President since May 2024.
The audit ran from May 2024 to December 2025 using a shared analytical framework of 9 topics and more than 92 structured questions, letting each SAI apply a common methodology to its own government while producing genuinely comparable, cross-country results — a 'parallel audit' rather than a single joint mission.
The consolidated findings describe a common pattern across the 12 countries: institutional ownership of AI initiatives is generally ahead of operational guidance, and ethics commitments often remain principles-level, without consistent, testable assurance mechanisms behind them. On infrastructure, participating governments are investing in AI compute capacity, but every country examined relies on third-party cloud providers even where national cloud environments exist, making hybrid delivery the practical norm rather than full sovereign control.
As an honest limitation: this is the audit's first cycle, so there is not yet evidence of the report changing government AI policy in any of the 12 countries, and the full methodological detail and country-by-country scoring were not independently re-verified outside the EUROSAI network for this write-up.
Read the full analysis: https://www.mevaker.gov.il/en/newsroom/2026-03-30
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.
Where this practice's information was retrieved from, and when.
United States of America
The IRS grew from 10 to 126 active AI use cases between 2022 and mid-2025 to flag high-risk tax returns …
Portugal
Portugal's supreme audit institution built a 37-indicator ML risk model to target procurement audits, developed with NOVA IMS, EU funding …
Belgium
Since 2014, Belgium's federal social security service has used the MiningWatch data-mining tool to rank construction, cleaning and hospitality employers …
Chile
ChileCompra cross-checks officials' asset declarations against suppliers' beneficial owners, flagging conflicts of interest before contracts are awarded. In 2025, flagged …
Open full copilot Grounded in cited practices — always check the sources.