OpenAI is calling on the UK government to introduce binding, mandatory safety rules for the most powerful AI systems, marking a significant shift toward formal regulation in a country that has relied on voluntary industry commitments. The company's top European lobbyist argues that legislation is now necessary given the rapid advancement of AI capabilities and recent safety incidents disclosed by leading AI firms.

Why Is OpenAI Backing Mandatory UK AI Legislation?

The push comes as the UK Labour government faces mounting pressure from lawmakers to address AI risks after companies including OpenAI and Anthropic disclosed safety incidents in recent months. The pressure intensified last week when a researcher at Anthropic resigned and publicly accused both Anthropic and OpenAI of "gambling with our lives." Despite Labour's 2024 manifesto commitment to "introducing binding regulation on the handful of companies developing the most powerful AI models," the government has since backed away from comprehensive AI legislation, instead relying on sectoral regulation and voluntary safety testing partnerships.

"The UK should act now to set up a stronger set of safety rules. OpenAI is in a position to support legislation in the UK to establish durable, mandatory, capability-based requirements for frontier AI that would build on the set of voluntary commitments that we currently have," said Tom Duff Gordon, OpenAI's head of policy in Europe, Middle East, and Africa (EMEA).

Tom Duff Gordon, Head of Policy in EMEA at OpenAI

Duff Gordon emphasized that the Labour government should "take advantage of this political window which is clearly opening up" to pursue legislation. He noted that such rules would put the UK on an "equal footing" with the European Union, which has already implemented the EU AI Act, and with emerging frameworks in the United States, where OpenAI has publicly backed "mandatory, capability-based national AI safety regulation".

Duff Gordon

What Would UK AI Safety Legislation Actually Look Like?

According to OpenAI's policy chief, any UK legislation should be narrowly focused on the most serious risks and the small number of labs developing frontier AI models. The framework should address national security and cybersecurity concerns rather than attempting to regulate all AI applications. Duff Gordon outlined several key components that he believes should be included in any new regulatory regime.

• Independent Testing Requirements: Compulsory third-party model testing before deployment to verify safety and performance across critical scenarios.
• Cybersecurity Focus: Strong emphasis on protecting AI systems from attacks and ensuring secure development practices across the industry.
• Incident Monitoring and Reporting: Requirements for AI developers to monitor, document, and report safety incidents to regulators in real time.
• Pre-Release Testing Standards: Flexible and adaptive technical rules for testing models before release, designed for international interoperability so the UK can align with emerging global consensus.
• Central Role for UK AI Security Institute: The government's AI Security Institute (AISI) should play a central role in the new regulatory regime, leveraging its technical expertise.

Duff Gordon stressed that legislation should be designed with "international interoperability" in mind, so that "as we move, we hope, towards greater global and multilateral consensus on the governance of frontier AI, the UK will be well positioned". This suggests OpenAI sees UK regulation as part of a broader effort to establish consistent global AI safety standards.

Duff Gordon

How Can the UK Establish Leadership in AI Safety Governance?

OpenAI argues that the UK has a unique opportunity to become a leader in what it calls "this new safety-conscious era." The company maintains its largest research hub outside the United States in the UK, giving it significant stakes in how the country regulates the technology. Duff Gordon emphasized that OpenAI has "high regard" for the technical competence and expertise of the UK's AI Security Institute, suggesting the company sees the institute as a credible partner in developing and enforcing safety standards.

The timing of OpenAI's push is notable. AI safety is expected to be on the agenda when US President Donald Trump and China's Xi Jinping meet later this month. Duff Gordon characterized this as "a very important conversation for them to have," but he also suggested there is an opportunity for the UK to have "a strong and influential voice on the next stage of policy" in global AI governance.

The call for binding UK legislation represents a notable position from OpenAI, which has historically emphasized the importance of industry self-regulation and voluntary commitments. The shift reflects growing recognition within leading AI companies that some form of mandatory oversight may be necessary to maintain public trust and address legitimate safety concerns as AI systems become more capable. Whether the UK government will heed OpenAI's call remains to be seen, but the company's endorsement of binding rules could provide political cover for lawmakers considering more stringent regulatory approaches.