TL;DR
OpenAI has asked the UK government to legislate now on frontier AI developers, itself included, while exempting smaller developers. Its head of European policy framed the moment as a political opening that ministers should use. The request lands five days after the company made a parallel pitch to Congress, and borrows almost exactly the wording of a promise Labour made in 2024 and has not kept.
What OpenAI is asking for
Tom Duff Gordon, who leads OpenAI’s policy work in Europe, said the company backs “stronger UK rules for the handful of companies, including OpenAI, developing the most powerful AI systems”. The scope matters as much as the endorsement. OpenAI wants obligations confined to frontier labs, with firms building less capable systems left outside.
Speaking separately to Politico, Duff Gordon urged the government to “take advantage of this political window which is clearly opening up”. That window was opened by Anthropic, whose chief executive called at the weekend for leading labs and governments to coordinate a slower pace of development. OpenAI, Google DeepMind and Elon Musk all endorsed the idea quickly.
Parliament is already pushing
The intervention coincided with the Joint Committee on Human Rights publishing its call for an AI Bill, a single regulator and some outright bans. Days earlier, more than 70 parliamentarians had asked Andy Burnham to support a ban on superintelligence. Downing Street declined, while saying it was looking at narrower measures on national security risks.
The government’s response to the committee left the door ajar. A spokesperson said the current framework should not be assumed to remain adequate as capabilities grow, and that “any future measures will be evidence-based”.
The gap OpenAI’s proposal would fill is a practical one. The AI Security Institute tests advanced models, but only with the developer’s cooperation. It was reported last week that Anthropic had not submitted its newest model.
Looking forward
The phrasing is the telling part. Labour’s 2024 manifesto pledged binding regulation on “the handful of companies developing the most powerful AI models”, and no bill has followed. OpenAI is now offering ministers their own commitment back, with industry support attached. A regime limited to a few frontier labs is also the version incumbents find easiest to live with, and that tension deserves scrutiny from MPs. For UK businesses building on these models rather than training them, a narrow frontier bill would add little direct compliance burden, though it would change what their suppliers must disclose.