How Governments Address Human Oversight of AI

An Overview of Domestic Regulations and International Frameworks

Report Image

Our report shows how governments address human oversight of AI at the domestic and international level.

Authors

Tommaso Giardini, Nils Deeg

Date Published

16 Nov 2026

Related Topics:

In June 2025, Anthropic tested 16 leading AI models: Each model read an inbox including information that an executive planned to shut it down that day and that he was having an affair. Most models decided to blackmail the executive. The test occurred in a fictional setup and Anthropic claimed not to have seen such behaviour in real-world deployment. But as AI systems become increasingly capable, a core concern for governments and firms is ensuring that humans retain control over AI decisions, particularly in high-risk contexts. The design of effective human oversight mechanisms is imperative, but complicated by the speed, scale, and opacity at which cutting-edge AI systems operate.

Our report shows how governments address human oversight of AI at the domestic and international level. It is part of our flagship report on The Fabric of AI Governance, which covers ten concerns.