Overview
- This week Parliament’s Joint Committee on Human Rights urged the government to bring forward an AI Bill that would put the AI Security Institute on a statutory footing and require pre‑deployment testing of high‑risk models.
- The committee recommended banning some uses of AI, forcing clearer disclosure when AI affects people, and requiring meaningful human oversight rather than a nominal “human in the loop.”
- Senior industry figures have publicly urged a slowdown in frontier model development and backed independent safety assessments, a stance that has pushed politicians to consider stronger regulation.
- On Capitol Hill lawmakers have opened bipartisan talks, proposed hearings and technical measures such as a government‑recognized “kill switch,” but support is uncertain because of presidential opposition, GOP leaders’ preference for a lighter approach, and an imminent congressional recess.
- If enacted the measures would force developers to submit models for testing, could favor well‑resourced firms able to meet costly compliance, and would reshape how companies deploy powerful AI that affects people’s rights and daily lives.