GAO Sets Framework for Accountable AI in Government
The Government Accountability Office has outlined how federal agencies should govern AI systems, with Sarah Harvey, the GAO's Director of Science and Technology Assessment, detailing the office's accountability framework in May 2026.
The framework rests on three pillars: data governance, security and transparency. It requires clear roles, multi-disciplinary stakeholder involvement, and strict controls over data handling and privacy.
Where agencies are using AI now
Federal agencies are already deploying AI across operations. The GAO uses AI in the Federal Audit Clearinghouse to analyze single audit data. The National Institutes of Health uses it for metadata tagging and image flagging. The Department of Veterans Affairs uses AI for medical imaging analysis.
These applications show both the potential and the risk. Each touches sensitive data or decisions affecting citizens.
What the framework requires
Governance means defining who owns AI decisions and who is accountable if something goes wrong. Data governance specifically addresses how information flows through systems and who can access it.
Security and privacy protections must be built in from the start, not added later. Transparency requires agencies to document how AI systems work and explain their outputs to stakeholders and the public.
The GAO's approach reflects a shift in how government thinks about AI deployment. Rather than asking whether an agency can use AI, the question is now how to use it responsibly.
For government professionals implementing AI systems, the framework provides a baseline. It signals that accountability is not optional and that agencies will face scrutiny on governance, data handling, and transparency.
Learn more about AI for Government and AI Data Analysis to understand how these principles apply to your agency's work.
Your membership also unlocks:
