OpenAI and Anthropic Ask UK Parliament: Make AI Safety Rules Binding

Two of the world's four frontier AI companies told British lawmakers on Tuesday that voluntary safety pledges no longer suffice and that Parliament should pass binding AI safety laws. The testimony from OpenAI and Anthropic, delivered to the Business, Innovation, Science, and Trade Committee, put an open industry split on the parliamentary record: Google DeepMind and Meta, appearing at the same hearing, declined to back new legislation.

The numbers frame the stakes. Four frontier AI firms gave evidence. Four former employees of those firms appeared as whistleblowers. Senior British civil servants testified in the final panel. And only a few hundred meters away, on the same day, MI5 director Ken McCallum used his annual threat update to name bioweapon misuse as the AI danger to which he attaches the greatest weight.

What did the companies actually ask for?

Anthropic went furthest. "The rules should be binding," Liam Booth-Smith, Anthropic's head of Europe, told the cross-party panel of MPs. He said the company has "consistently believed that there should be binding rules on companies and that voluntary commitments are not enough."

OpenAI took a more cautious line but reached the same destination. Tom Duff Gordon, the company's vice president, backed a statutory framework and an international convention. "There should be some elements that should be mandatory," he said, naming incident reporting as the starting point. His verdict on the status quo was blunt: "We don't think basing everything on voluntary commitments for frontier risk is a good long-term plan."

The rivals refused to follow. Asked directly whether he supported a statutory framework, Google DeepMind's Owen Larter said: "Focus on the standards is what we would be doing." Meta's Rob Sherman pointed to Britain's existing regulators and told MPs they were "not writing on a blank page."

The divide matters because the United Kingdom, under its current arrangements, has no statutory lever over frontier model releases. The hearing demonstrated, in the government's own words, that the entire regime rests on corporate goodwill — a point civil servants conceded under questioning later in the session.

Why did the committee stage this showdown?

Committee chair Liam Byrne told WIRED after the hearing that exposing the divide in a second jurisdiction was deliberate. The committee set out "to light up the wide differences in opinions between some of the wealthiest firms on Earth," he said, and to show "that you couldn't rely on voluntary self-regulation."

Byrne then linked the parliamentary testimony directly to the security warning delivered the same day. "The fact that the head of MI5 is warning about incredibly dangerous capability uplifts for bioterrorists in one part of Whitehall, and then in another part of Whitehall the permanent secretary of the Cabinet Office is basically admitting that our regulatory regime for AI is voluntary, tells you everything you need to know," he told WIRED.

McCallum's own words carried weight in that framing. "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," the MI5 director said in his annual threat update.

How voluntary is the current British regime?

Entirely, according to the official responsible for it. Britain's AI Security Institute (AISI), the body that tests frontier models before release, only examines systems when companies choose to hand them over. Henry de Zoete, the institute's director, acknowledged that AISI depends wholly on the cooperation of companies and foreign governments for access.

Cat Little, the top civil servant at the Cabinet Office, which houses AISI, was pressed on whether the whole regime was voluntary. "At the moment, yes, it is," she said.

Pressed further on what would happen if AISI judged a model too dangerous and a company released it anyway, de Zoete conceded the institute has no power to prevent the release. "What we're not doing is kite marking," he said, using the British term for an official safety certification. "We're not saying yes or no to any sort of model."

De Zoete also located the limit of British leverage geographically. "These companies are either American or Chinese who are building at the frontier … There's only so much we, as the UK, can do in that kind of context," he said.

Where does Washington fit in?

The executives made explicit that British access to frontier models runs through the US government. "We provide our models predeployment to CAISI," OpenAI's Duff Gordon said, referring to AISI's American counterpart. "We will continue to provide our models pre-deployment to AISI in the UK … and as long as we're able to do that under the US government, we will continue to do it."

Anthropic's Booth-Smith confirmed the constraint with a concrete case. "We unfortunately were not able to provide full unrestricted access to [Mythos 5.1] to UK AISI at that time ... That's a question for the US government," he said, before adding that AISI now has access to that model.

The exchange establishes a fact with policy weight: a US administration can effectively gate what British safety testers see, regardless of what Westminster wants.

What did the whistleblowers say?

Four former lab researchers testified, and their testimony aligned with the pro-legislation camp. Daniel Kokotajlo, a former OpenAI staffer, told MPs: "I don't think the companies can be trusted to self-regulate."

Steven Adler, who left OpenAI in late 2024, said he was "not confident in the safety culture across the industry."

Geoffrey Irving described the labs' internal logic: the companies believe they are locked in a race that governments will not interrupt and that they cannot leave. "Both of those are false," he said. "They could just stop."

What happens next?

Byrne told WIRED his committee will now spend significant time examining the United Kingdom's path forward as a middle power in AI. His specific proposal: ministers should use Britain's coming G20 presidency, which South Korea follows in the chair, to broker an international convention that does not depend on the current White House. "The G20 presidency will outlive the US presidency of Trump," he said.

He closed with a warning that frames the legislative question now before Parliament: "The risks are multiplying faster than our safety regime can control."

With two frontier firms now on record asking for binding law, two refusing it, the MI5 director naming bioweapon uplift as his top concern, and the Cabinet Office's own permanent secretary confirming the regime is voluntary, the debate has moved from whether Britain legislates to what a statutory framework would compel — and whether any British rule can bind labs whose releases are gated in Washington.