NOISESaved copy
Coverage history

Assembled by NOISE.

Saved report copy

Recorded

OpenAI and Anthropic urge UK lawmakers to impose binding AI rules

Representatives for OpenAI and Anthropic told a British parliamentary committee on Tuesday that voluntary commitments are not enough and that AI safety should rest on binding law, with Anthropic's Liam Booth-Smith saying rules should be statutory and OpenAI's Tom Duff Gordon backing mandatory incident reporting. Google DeepMind and Meta declined to back new legislation, exposing a split among frontier firms, and civil servants conceded the UK regime, including the AI Security Institute's pre-release testing, is entirely voluntary. The hearing came as MI5 director Ken McCallum warned AI could change the stakes on bioweapons, and committee chair Liam Byrne said he wants Britain to use its G20 presidency to broker an international convention. The role reversal matters because the leading labs are the ones inviting binding oversight terms, while Anthropic disclosed that UK access to its latest model depended on US government approval.