Labour conference set to vote on new AI regulator as cabinet concerns grow

Proposals to create a more powerful regulator for AI in Britain will be voted on at Labour conference next week, as concerns mount in Westminster about the speed with which the technology is transforming society and business.

The looming vote, which has been proposed by the Communication Workers Union and seeks to replace the AI Security Institute (Aisi), comes as the AI minister said the UK government would consider strengthening legislation if necessary.

Louise Haigh, first secretary of state, told the Trades Union Congress annual conference this week that AI posed “huge risks” to UK national security without proper protections and that ministers had to “heed the warnings” of those on the cutting edge of the technology.

Another senior minister told the FT that AI was being raised more and more at weekly cabinet meetings in Downing Street. “I think in a year’s time AI will dominate all of our deliberations and discussions, it’s the biggest issue in almost every area,” the person said.

Beyond Westminster, AI is rising up the agenda in international forums. While the wars in Ukraine and the Middle East dominate multilateral talks between foreign ministers, AI is fast becoming the third most important issue, according to diplomats.

Fears about the potential for AI to cause harm have surged up the agenda after current and former staff in the fast-evolving sector warned that the technology could destroy humanity.

The FT revealed this month that Anthropic declined to submit its latest model to the Aisi for testing before release, in a blow to the global leader in the testing of frontier AI models.

While US President Donald Trump rejected calls for an AI slowdown following the dramatic safety warnings, King Charles this week warned of “existential dangers” at a summit of tech bosses to discuss the future of the technology.

But ministers are also grappling with more immediate consequences of AI’s rollout, including its major disruption to labour markets as jobs are replaced by computers.

Mike Clancy, general secretary of the Prospect trade union, which represents about 157,000 professionals, said that “there isn’t a meeting we go to with government where they don’t talk about AI”.

The challenge for the government, he added, was that it wanted the UK “to be a very positive space for AI development and to be known for that, while also avoiding being handmaidens of a jobless future”.

The Labour conference motion put forward by the Communication Workers Union, which represents 7,000 workers in the tech sector, including some staff at Google DeepMind, will call for a regulatory overhaul in the sector. The vote is non-binding.

John Chadfield, CWU national officer for tech workers, said the Aisi was under-resourced, understaffed and toothless in terms of its powers.

“It’s completely opt-in for those big companies who have got very deep pockets and good influence with governments around the world,” he said. “It’s not a regulator. Companies don’t have to put anything in front of it.”

Liam Byrne, Labour chair of the House of Commons business committee, this week called on Aisi chief Henry de Zoete to give evidence to MPs. In a letter, Byrne said he had serious concerns about “the adequacy of current AI safety governance”.

Chadfield said he did not oppose the Aisi, which was set up three years ago, but that it should be one piece of a wider framework of legislation, enforcement agencies and regulatory bodies.

He said: “You can’t both have it [AI] be the economic miracle that’s going to be the backbone of the country and the scariest thing to happen in decades and have zero regulation. Make it make sense.”

Darren Jones, who was first secretary to former prime minister Sir Keir Starmer, said it was right to respond to calls to “help pace the frontier” of AI.

Jones stressed that Aisi was a “world-leading team” for AI safety but added that the government should “offer to scale up Aisi to provide its services to other countries around the world as part of a new multilateral framework of the oversight of so-called ‘super-intelligence’.”

Other union leaders are pressing the government to begin developing policies to protect workers if AI starts to wipe out jobs at speed and scale.

Sharon Graham, general secretary of Unite, said AI models trained on workers’ data were mimicking them “without anything being given back” and pointed to proposals to launch a “robot tax” in New York as a potential model.

“The bullet has to be bitten” on the question of how to tax AI, she added.

The threat from AI loomed large at the annual TUC conference in Brighton, where delegates passed a motion urging the government to do more to “keep pace” with the deployment of the new technology.

The motion recognised some potential benefits from AI but warned of negative consequences including “deskilling, job displacement, increased surveillance, work intensification, exploitation and widening inequality”.

Kanishka Narayan, AI minister, told The Times this week Britain was prepared to introduce new regulations and legislation if they were needed.

“There is no credible choice between being ambitious about AI and being serious about its risks. We must do both,” he said. “Nothing is off the table. Where further regulation, legislation, stronger protections or new powers are needed, we will act. The test I will apply is simple: will it make the British people safer?”