OpenAI backs binding UK AI legislation
LONDON — OpenAI’s top European lobbyist is calling for the U.K. to introduce binding rules on AI developers in light of the technology’s rapidly advancing capabilities, and the risks that come with them.
Tom Duff Gordon, the company’s head of policy in EMEA, told POLITICO in an interview that the “U.K. should act now to set up a stronger set of safety rules.”
“OpenAI is in a position to support legislation in the U.K. to establish durable, mandatory, capability-based requirements for frontier AI that would build on the set of voluntary commitments that we currently have,” said Duff Gordon.
He said Andy Burnham’s government should “take advantage of this political window which is clearly opening up” to pursue legislation.
The Labour government is facing mounting pressure from lawmakers to address the risks of AI after leading companies disclosed a spate of safety incidents in recent months. That pressure intensified last week after a researcher at AI company Anthropic resigned and accused Anthropic and OpenAI, where he’d previously worked, of “gambling with our lives.”
In its 2024 manifesto, Labour committed to “introducing binding regulation on the handful of companies developing the most powerful AI models,” but has since backed off the idea of comprehensive AI legislation, instead relying on sectoral regulation and safety testing by its AI Security Institute through voluntary partnerships with AI firms.
Duff Gordon said legislation would put the U.K. on an “equal footing” with the EU and emerging state and federal frameworks in the U.S. — where OpenAI has said it backs “mandatory, capability-based national AI safety regulation.”
But he stressed that any U.K. legislation should be narrowly focused on the most serious risks related to national security and cybersecurity, and “the small number of labs and the small number of models that are completely at the frontier.”
He said any legislation should include compulsory independent model testing by third parties, a “strong focus” on cybersecurity, and requirements for monitoring and reporting safety incidents. AISI should play a “central role” in any new regime, he added, saying OpenAI has “high regard for their technical competence and expertise.”
Legislation should also include standards for testing models ahead of their release set out in “flexible and adaptive” technical rules designed for “international interoperability,” he said, so that “as we move, we hope, towards greater global and multilateral consensus on the governance of frontier AI, the U.K. will be well positioned.”
AI safety is expected to be on the agenda when U.S. President Donald Trump and China’s Xi Jinping meet later this month, which Duff Gordon said was “a very important conversation for them to have, and we hope that they have it.”
But he said there is also an opportunity for the U.K. — where OpenAI maintains its largest research hub outside of the U.S. — to have “a strong and influential voice on the next stage of policy.”
“We think the U.K. is a leader in AI,” Duff Gordon said. “We want them to continue to be a leader in this new safety-conscious era that we’re going to be moving into.”
