Representatives from OpenAI and Anthropic called for binding, statutory AI safety regulations during a UK parliamentary hearing, creating a stark divide with industry peers Google DeepMind and Meta. Appearing before the Business, Innovation, Science, and Trade Committee, executives from Anthropic and OpenAI argued that voluntary commitments are insufficient for managing frontier risks, urging lawmakers to mandate incident reporting and safety standards.

In contrast, executives from Google DeepMind and Meta advocated for relying on existing regulatory bodies and voluntary industry frameworks rather than passing new binding legislation. The hearing highlighted growing political interest in formal oversight, with committee leaders noting the contrast between voluntary government testing regimes and security warnings regarding AI-enabled bioweapon threats.

Under current UK rules, the AI Security Institute relies entirely on voluntary cooperation from AI companies to inspect frontier models prior to deployment, a structure parliamentary leaders indicated may need reform.

Why it matters

  • Regulatory divergence among top AI labs increases the likelihood of mandatory compliance frameworks in the UK and European markets.

  • Frontier AI startups should prepare for mandatory incident reporting and formal safety audits as voluntary guidelines face political backlash.

  • Enterprise buyers may favor labs advocating for statutory compliance to minimize future regulatory legal liability.

Source: wired.com