OpenAI and Anthropic's message to British lawmakers this week: don't trust us, regulate us.
Executives from both companies appeared at a sweeping UK parliamentary hearing alongside AI safety whistleblowers and security researchers. They urged MPs to put binding safety rules into law rather than rely on company promises. The hearing doubled as a venue for broader AI anxiety: a few hundred meters away, Britain's spy chief warned that AI could help someone build a new bioweapon.
When the companies building the most powerful AI systems ask to be regulated, it's worth asking why. Binding rules are easier to predict than voluntary pledges that shift with every model release, and they can also raise the compliance bar high enough to keep smaller rivals out.
It's a familiar move: ask to be regulated, then help write the rules that end up protecting incumbents most.