Representatives from OpenAI and Anthropic have urged British lawmakers to introduce new laws to regulate AI safety, contrasting with Google DeepMind and Meta, which are holding back. The Business, Innovation, Science, and Trade Committee held a hearing with executives from four AI firms, and former staff turned whistleblowers, highlighting the divide in approach.

‘The rules should be binding,’ Anthropic’s Liam Booth-Smith declared, emphasizing the need for statutory frameworks. OpenAI’s Tom Duff Gordon was more cautious, supporting a statutory framework and an international convention, but still insisting on mandatory elements like incident reporting. In contrast, Google DeepMind’s Owen Larter focused on standards, while Meta’s Rob Sherman pointed to existing regulators.

The hearing also revealed Britain’s security service director, MI5’s Ken McCallum, warning of the bioweapon misuse threat posed by AI. This warning, coupled with the voluntary nature of current British regulatory regimes, sparked a debate on the adequacy of self-regulation. Civil servants admitted that Britain’s AI Security Institute (AISI) relies heavily on voluntary cooperation from companies and foreign governments.

Executives also discussed the limitations of British access to AI models, with OpenAI’s Duff Gordon stating that pre-deployment access to AISI is contingent on the US government. The tension between self-regulation and statutory frameworks was evident, with the committee chair, Liam Byrne, emphasizing the need for action.

Following the hearing, Byrne announced that his committee would dedicate time to examining Britain’s path forward as a middle power in AI, aiming to broker an international convention during the country’s G20 presidency.