OpenAI and Anthropic Want AI Laws — Google and Meta Don't

OpenAI and Anthropic, two of the world's biggest AI companies, told a British parliamentary committee that voluntary safety pledges are no longer enough. They want binding laws to ensure AI doesn't harm people. "The rules should be binding," said Anthropic's Europe head Liam Booth-Smith. OpenAI's Tom Duff Gordon agreed, saying some things, like reporting AI accidents, should be mandatory. But their rivals refused to back new laws. Google DeepMind said it prefers focusing on standards, and Meta said Britain already has regulators. This public split shows even AI leaders can't agree on how to keep you safe.

The hearing also revealed how weak the current system is. Britain's AI Security Institute (AISI) tests new AI models before release, but only when companies choose to hand them over. Its director, Henry de Zoete, admitted AISI has no power to stop a dangerous model from being released. "At the moment, yes, it is" voluntary, said Cat Little, the top civil servant at the Cabinet Office. This means if an AI company decides to release a risky model, the UK government can't block it. That's a problem because AI is getting more powerful and could be misused.

Just a few hundred meters away, MI5's director Ken McCallum warned that AI could help terrorists create bioweapons. "Not yet realized, but the misuse case to which I'd attach greatest weight, is AI's potential to change the stakes on existing or even new bioweapons," he said. This stark warning highlights why some companies want laws. But without agreement, the UK may struggle to protect its citizens. The committee chair, Liam Byrne, said the hearing showed "you couldn't rely on voluntary self-regulation." The divide between AI companies means progress on safety rules may be slow, leaving you and your family vulnerable to potential AI risks.