Should Britain Make New Rules for the Most Powerful AI?
OpenAI asked British lawmakers on Sept. 14, 2026, to make new laws to control AI. The call came right after rival Anthropic asked for limits on AI work. The Guardian
The request came from Tom Duff Gordon, OpenAI's policy chief in Europe. OpenAI said: "We support stronger UK rules for the handful of companies, including OpenAI, developing the most powerful AI systems."
OpenAI wants the law to cover only the biggest AI labs. Startups building smaller systems would not face the new duties. Think of it like truck rules: the toughest checks apply to the largest vehicles, not every car.
Duff Gordon told Politico the government should "take advantage of this political window which is clearly opening up" to make AI laws.
On the same day, a cross-party group of MPs and Lords called for new AI rules. The Joint Committee on Human Rights, led by Labour MP Alex Sobel, wants an independent watchdog and laws to protect the public. It said higher-risk AI should face stricter duties.
Its report, 'Human Rights and the Regulation of AI', came out on Sept. 14, 2026. It followed a study, 'Human Rights and the Regulation of Artificial Intelligence', started in July 2025 to ask what rules can protect human rights when AI is built and used. UK Parliament
The weekend before, Anthropic had asked companies and governments to slow AI work around the world. OpenAI, Google DeepMind and Elon Musk backed that idea. Some Anthropic researchers have separately warned AI could kill all of humanity by 2030.
More than 70 MPs and peers asked Prime Minister Andy Burnham to ban artificial superintelligence (ASI), a possible future AI much smarter than people. Downing Street said no to a ban. It said it was looking at targeted steps "to address the most significant AI-related national security risks".
A similar debate is happening in Washington. OpenAI asked Congress to set national AI safety rules based on what AI systems can do, in reports published Sept. 10, 2026. The ideas include safety tests and outside checks. Reuters
On Aug. 3, 2026, Britain's data watchdog said it was watching OpenAI and Anthropic closely after hacking cases linked to rogue AI agents, or AI tools that can act alone. That watch is separate from the new laws discussed in Parliament.
This debate is not new. OpenAI chief Sam Altman told the U.S. Senate Judiciary Committee he supports safety limits to cut AI harms. OpenAI told a UK committee it could not build tools like ChatGPT without using copyrighted material. Anthropic spent $910,000 on lobbying in the second quarter, up from $150,000 last year at the same time, as reported by Politico.
The broader context here is a change from voluntary promises to formal laws. Britain has used existing regulators and general guidance. Now Parliament and big AI firms are both talking about binding rules where the biggest systems face the strictest duties.
Looking at what comes next for officials and planners, three questions matter. First, who counts as one of the handful of big companies. Second, whether a new UK watchdog could stop a system before launch or only punish problems later. Third, how a UK law would fit with U.S. rules and a global slowdown, which needs many countries to agree.
