OpenAI and Anthropic Urge UK Lawmakers for Binding AI Regulation

Representatives for artificial intelligence companies OpenAI and Anthropic have urged the British Parliament to establish a statutory regulatory framework for AI safety. Speaking at a parliamentary hearing on Tuesday, the companies' delegates stated that voluntary pledges are insufficient and new legislation is necessary.

Liam Booth-Smith, head of Europe for Anthropic, affirmed his company's consistent belief in the necessity of binding rules. Tom Duff Gordon, vice president at OpenAI, supported a statutory framework and an international convention, suggesting mandatory elements such as incident reporting. Duff Gordon argued against relying solely on voluntary commitments for managing frontier risks.

In contrast, competitors Google DeepMind and Meta did not endorse new legislation. Owen Larter, representing Google DeepMind, indicated a focus on standards, while Meta's Rob Sherman pointed to the UK's existing regulatory bodies. This divergence during the hearing highlighted internal disagreements within the AI sector regarding the need and nature of regulation.

The hearing occurred as the head of the UK's security service, MI5, warned of potential AI-driven dangers, including enhanced biothreats. MI5 Director Ken McCallum cited AI's potential to alter the stakes for biological weapons as a significant, though not yet realized, risk. An official from the UK's AI Security Institute acknowledged that the current system does not compel companies to submit models for testing.