Topics on AI Governance

Organizations are shifting AI governance from policy documents to operational controls: living inventories, risk triage, audit evidence, logging, access restrictions, and continuous monitoring. Agentic AI and shadow AI are driving much of the current pressure.

Recent reporting is dominated by a delayed Trump White House executive order on frontier AI oversight. The draft would let federal agencies review advanced models before release, but the plan remains unsettled amid disputes over voluntary access, security testing, and fears of de facto licensing. National security agencies, major AI labs, and White House advisers are the main actors, while the policy direction still looks fluid and unresolved.

Universities are tightening student AI rules, especially for exams and assessed writing, with Berkeley Law and Edinburgh Philosophy both moving toward stricter default bans and narrower allowed uses. The pattern favors integrity, disclosure, and limited instructor discretion over broad student access.

USDA watchdog reports show the department is still missing core AI governance and cybersecurity controls, including required OMB policies, reliable inventory tracking, and formal risk reviews before deployment.

Researchers at Mount Sinai created a new index to map health care AI policy across jurisdictions and institutions. The underlying finding is that oversight is expanding quickly, but remains fragmented, leaving health systems without a single unified framework for deployment, monitoring, and governance. The work emphasizes practical governance needs such as transparency, patient safety, and accountability.