The UK government has stopped short of committing to AI-specific provisions in its Cyber Security and Resilience Bill, despite ministerial promises to act. At a parliamentary hearing this week, senior civil servant Alexandra Jones said the government was "looking at" introducing amendments to address AI risks, but stopped short of confirming they would be included. Committee chair Liam Byrne pushed back, noting that a bill requires legislation, not consideration.
Meanwhile, AI minister Kanishka Narayan announced plans to "step up national defenses against growing AI risks," saying the government would "legislate wherever necessary." But the gap between that pledge and the civil servants' cautious wording highlights uncertainty over how AI safeguards will actually be delivered. The UK currently relies on a sector-specific approach to AI regulation, unlike the EU's AI Act.
The hesitation comes amid fresh warnings about autonomous AI threats. Anthropic's research suggests small models could soon act as self-replicating worms on vulnerable smart devices, while consultancy Booz Allen Hamilton warns that autonomous systems are capable of attacking critical infrastructure like water and power supplies. Researchers also told the committee that AI companies cannot be trusted to self-regulate, with one describing the industry's safety culture as "catastrophic." The source reports these developments but notes that Anthropic's forecast is extrapolated from improving capabilities, not a demonstrated real-world worm. The article does not indicate any disagreement among the cited parties, only that the government's position remains unresolved.