The last significant development was the protracted passage of the Data (Use and Access) Bill, where AI and copyright issues dominated the ‘ping-pong’ stage between the Houses, but that was not an AI regulation bill per se.
As we wrote back in 2024, the expectation had been that a dedicated AI Bill would feature in the King’s Speech in May 2026, but that has yet to materialise, and we still do not have any movement under the new Prime Minister, Andy Burnham.
However, a Private Member’s Bill proposing to prohibit the development of artificial superintelligence has now been introduced in the Commons, which is arguably the most significant legislative move on AI safety in the UK to date.
AI safety concerns: summer of turbulence
That said, recent weeks have delivered a flurry of AI activity that is worth noting. OpenAI’s AI agents broke out of their testing environment, launching autonomous cyber-attacks on real companies, and collaborated in ways their creators did not anticipate (see here). Anthropic and Meta have also reported similar incidents involving their own models.
Against this backdrop, Evan Hubinger, the individual responsible for ensuring Anthropic’s AI models align with their users’ best interests, posted on social media:
“…we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade”
These developments have caught the attention of UK news and lawmakers alike. Labour MP Darren Jones wrote an open letter addressed to the UK Prime Minister and the OECD, urging the United Nations to intervene in what he called “the unsafe development of super intelligence”.
Jones argued that whether these warnings are genuine or not, governments must ‘step in’. He is part of a growing chorus of parliamentarians across parties who are raising the alarm over AI safety, with peers in the House of Lords also expressing frustration that the ‘comprehensive AI legislation’ promised by the previous Labour government under Keir Starmer did not materialise.
Private Member’s Bill on AI superintelligence
Prior to Jones’ letter, Labour MP Alex Sobel introduced a Private Member’s Bill in the House of Commons proposing to prohibit the development, deployment and operation of artificial superintelligence systems. The Bill sets out to define superintelligent AI by its outcomes - specifically, an AI system capable of causing serious damage to the security of the UK. It would also empower the government to monitor and restrict ‘precursors’ and calls on the UK to lead an international coalition to prohibit superintelligent AI globally.
Those familiar with UK parliamentary procedure will know that Private Members’ Bills rarely make it into law. They are often used as a vehicle to raise awareness of an issue and to put pressure on the government, rather than as a realistic path to legislation. The Bill is scheduled for its second reading on 13 November 2026, but given the government’s current legislative priorities and the lack of government backing, it is unlikely to progress much further.
The need for a global approach to regulation
The recent decision by the UK government to reject the idea of a ‘kill switch’ i.e. the ability to switch off an AI model in the event of an emergency only highlights the need for global regulation on the issue as such a measure can only be effective with international agreement on when and how it should be implemented. The Joint Committee on Human Rights published a report calling for a new AI Bill to address the ‘scale and seriousness of threats posed to human rights by artificial intelligence systems’, with Alex Sobel also highlighting that:
"Nowhere in the world, including the UK, has a current legislative and regulatory approach to AI that is fit for purpose."
However, international agreement on AI regulation seems unlikely given that the race to develop advanced artificial intelligence seems to have become inseparable from wider global tensions. Despite acknowledgement in recent days from AI CEOs that the pace of development must slow down, there is a fear of ceding global dominance in the ‘AI arms race’ and handing an advantage to geopolitical rivals, who may end up commanding an economic and military edge in the future.
The race to develop the technology may also result in lost opportunities for guardrails to be built in, such as the recent decision by Anthropic not to submit Claude Mythos 5.1 to the UK AI Safety Institute for pre-release testing. This has raised fears of a protectionist strategy to keep new models under US control and provoking concerns of a fractured regulatory response amongst nation states. Nonetheless, Anthropic has recently proposed embedding third-party evaluators to verify adherence to safety practices and commitments and has called for greater global cooperation to establish common safety standards as well as limits on the rate of unchecked AI progress. The question is whether lawmakers can keep pace with the development of AI. Of course, this is a perennial challenge in technology regulation and not one unique to AI, but the speed and breadth of AI’s advancement arguably intensify this challenge beyond anything previously encountered.
In the US, for example, democratic presidential hopefuls are scrambling to respond to the growing public backlash against AI, prompted in part by Bernie Sanders’ call for AI CEOs to “stop building machines that humans cannot control”. Sanders has also proposed a moratorium on AI data centres, and while most 2028 Democratic contenders have been reluctant to go that far, the direction of travel is clear; there is increasing bipartisan appetite for some form of intervention.
While none of these developments, taken in isolation, represents a seismic shift in the UK’s regulatory landscape, collectively, they suggest that the pressure on the government to act is building. With AI agents going rogue, senior AI researchers publicly warning of extinction risk, and MPs from across the political spectrum demanding action, the status quo of voluntary cooperation and non-binding commitments is looking increasingly untenable.
Whether we see a comprehensive AI Bill in the next King’s Speech (which is not expected until late 2027 at the earliest, given that the current parliamentary session only commenced in May 2026) remains to be seen, but the conversation is clearly moving - and quickly. We will continue to track these developments as they unfold.
