GAO Issues AI Accountability Framework — June 30, 2021
The U.S. Government Accountability Office released an AI accountability framework guiding federal agencies on governance, data quality, performance, and monitoring controls.
GAO-21-519SP translates audit practices into an AI lifecycle framework covering four principles—Governance, Data, Performance, and Monitoring—and associated auditing questions. It emphasises documenting responsibilities, ensuring data integrity, testing models, and tracking outcomes for equity and mission effectiveness.
- Governance. Agencies must define roles, risk tolerance, and oversight mechanisms.
- Data. Teams should document provenance, representativeness, and privacy safeguards.
- Performance and monitoring. The framework promotes pre-deployment testing, continuous evaluation, and incident response planning.
The GAO guidance aligns closely with NIST’s AI RMF and supports Zeph Tech’s federal assurance engagements.