Security / AI Governance / Data Quality
GAO sets out 16 policy options on AI and cybersecurity, flagging data quality and governance gaps
The US audit office says AI magnifies attacks and aids defenders, but experts it consulted cited data privacy and quality, outdated governance frameworks and weak data sharing as obstacles.

The US Government Accountability Office (GAO) has published a report, AI and Cybersecurity: Technologies Can Magnify Cyberattacks and Enhance Cyber Defense (GAO-27-108458). It finds that AI helps threat actors produce phishing and deepfake content and that agentic AI can run several attack stages with little human input. One case it cites is a June 2026 incident in which multiple AI agents tested malicious code against security products.
GAO reports that there are no widely accepted criteria for evaluating generative AI’s offensive or defensive capabilities. Experts it consulted named data privacy and quality, outdated governance frameworks, fragmented regulation and insufficient data sharing as challenges.
GAO sets out four policy goals and 16 policy options, which it describes as a non-exhaustive menu for legislators, agencies, standards bodies and industry, not as requirements. According to CDO Magazine, the options include stronger AI testing, agent identity management and human authorisation for consequential AI-driven security actions. GAO published the report on 20 November 2026.
Why it matters
The report treats data quality, privacy and data sharing as limits on how well AI security tools work, so governance teams should be able to show what data feeds those tools and who is responsible for it. Its goal of governing AI tools across their cybersecurity lifecycle points to concrete controls: validating outputs, defining fallback procedures if an AI system fails, and assigning accountability for consequential actions. Although the options are not binding, they signal where US agencies, standards bodies and procurement requirements may move.
Status: Published. Policy options are a menu for policymakers, not mandatory requirements
Sources
Was this useful?
Thanks. What was the problem?
Thank you — this helps us choose what to cover. If you spotted an error, please tell us what it is.
Prepared with AI assistance and reviewed by the editor before publication. Spotted an error? Tell us — see our editorial policy and methodology.
Governance areas
Who should care
- CISO
- AI governance lead
- CDO
- Data stewards