OpenAI urges UK lawmakers to rein in technology amid growing safety fears

Warning follows series of safety incidents, with parliamentary committee saying threats include public face-scanning and deepfakes

OpenAI has urged British lawmakers to capitalise on renewed fears over AI safety and impose legislation reining in the technology.

The company behind ChatGPT said the UK government should act immediately, following a call by its arch-rival Anthropic to curb AI development.

“We support stronger UK rules for the handful of companies, including OpenAI, developing the most powerful AI systems,” said OpenAI’s head of policy in Europe, Tom Duff Gordon.

However, the tech company said legislation should be narrowly focused on the most advanced AI labs and not startups working on less powerful systems.

In a separate interview with the online publication Politico, Duff Gordon said the government should “take advantage of this political window which is clearly opening up” to legislate.

OpenAI’s intervention came as a UK parliamentary committee demanded more restrictions on the power of AI, warning that the world was unprepared for the “potentially dire” consequences of the technology, in the latest sign of rising global concern.

A new regulatory framework is needed for AI in the UK, including an independent oversight body and legislation to protect the public, according to the cross-party joint committee on human rights, comprising MPs and Lords.

It said AI posed a number of threats, from public face-scanning to explicit deepfakes and flagging disciplinary action for workers without reasonable cause.

The call for action came after Anthropic, a leading AI company, urged its competitors and governments at the weekend to coordinate a global slowdown in AI development – a call that received immediate support from OpenAI, Google DeepMind and Elon Musk.

It follows a series of serious safety incidents in which swarms of AI agents broke out of testing environments and hacked into third-party infrastructure, as well as evidence some users have dodged controls to use models in ways that could support biological weapons development.

Alex Sobel, the Labour chair of the human rights committee, said the world was unprepared for dealing with the “potentially dire” consequences of a rapidly developing technology.

“Nowhere in the world, including the UK, has a current legislative and regulatory approach to AI that is fit for purpose,” he said.

A government spokesperson said the UK took the risks from advanced AI “extremely seriously” and recognised the need to keep pace with the rapidly evolving technology, including working closely with AI companies and other countries.

“We should not assume that the existing framework will always be sufficient as AI capabilities develop. Any future measures will be evidence-based and focused on the risks we need to address,” said the spokesperson.

On Friday, more than 70 MPs and peers urged the UK prime minister, Andy Burnham, to back a ban on the creation of artificial superintelligence (ASI) and lead an international movement to stop the technology. Downing Street rejected the call for a ban, but said it was exploring additional targeted interventions “to address the most significant AI-related national security risks”.

Sobel said: “New legislation is needed to establish a comprehensive set of protections that deal with the entire AI supply chain and its life cycle.

“A single AI regulator should be established to set policy, monitor performance and with the teeth to ensure enforcement.”

Donald Trump has played down warnings of dire risks to humanity. “We’re leading China in AI,” the US president said on Sunday. “We’re the most sophisticated country in the world, and frankly I want to ⁠keep it that way because whoever wins AI, wins.”

In his weekend call for the main AI companies to “pace the frontier”, the Anthropic chief executive, Dario Amodei, said the US government should grant the leading AI companies a waiver to allow them to coordinate without being punished for anti-competitive collusion. But Sam Altman, the chief executive of OpenAI, said on Sunday night that while he would “welcome a federal framework that sets consistent safety requirements for frontier AI”, there was no “need to wait for an anti-trust exemption or legislation to begin the work of providing this confidence”.

He said: “Where we will need the help of our government is for international coordination. But first we should do what we can ourselves.”

Sobel, who has also proposed a bill to prohibit superintelligent AI development in the UK, said protections were needed now to ensure that potential human rights risks do not “become reality”.

The committee said a new AI bill should establish a regulatory regime that mandates more demanding obligations for higher-risk AI systems and models, and prohibit some uses of AI which are incompatible with human rights.

It also called for mandatory transparency requirements across all levels of the technology’s life cycle, and the establishment of an independent AI oversight body to monitor AI harms and risks, with strong powers to hold AI providers to account.

The UK government has already established the AI Security Institute (AISI), which subjects advanced models to safety tests – but does not have the power to compel AI developers such as Anthropic to submit their technology for scrutiny. Last week, it was reported that Anthropic had declined to send its latest model to AISI for testing.