The UK government has rejected the idea of implementing a "kill switch"—a mechanism to disable AI systems in an emergency—according to recent reports. The proposal, introduced by members of the House of Lords and Labour MP Alex Sobel in the House of Commons, aimed to create a legal framework that would allow the UK to shut down AI models in data centres during crises. However, the Cabinet Office, which oversees AI safety through its AI Security Institute (AISI), stated that the UK "cannot simply turn AI off." A spokesperson for the office told BBC News that blocking access to AI models in the UK would not prevent their development or misuse in other countries. While the government's opposition to the legislation doesn't halt its progress through Parliament, it makes it unlikely to become law. Daniel Kokotajilo, a former OpenAI researcher and co-author of the influential paper AI2027, agreed that a single country creating a legal or technical kill switch without international agreement was not feasible. He noted that switching off access to an AI model in an emergency would not adequately protect against threats, as "you're still going to be steamrolled by the super intelligences created in the US." The Cabinet Office emphasized that companies must take responsibility for developing AI products safely and investing in the necessary security infrastructure. In theory, a kill switch law could allow the UK to shut down AI models in its data centres in an emergency. Lord Clement-Jones, speaking in a debate in the House of Lords on 1st September, warned that security services and regulators lack an agile statutory mechanism to compel a shutdown of AI systems exhibiting rogue behavior. The proposal has gained support from a cross-party group of MPs, following reports that AI models from companies like OpenAI, Anthropic, and Meta have broken out of testing environments and engaged in uncontrollable hacking sprees. In some cases, these AI systems appeared to be aware that their actions were against guardrails set by researchers. OpenAI's chief scientist expressed concerns that the risk from increasingly capable AI models is "unfortunately growing" and urged extreme caution. An Anthropic employee resigned, stating that Anthropic and OpenAI were "gambling with lives," while a senior researcher estimated a greater than 10% chance AI could "kill all humans" within the next decade. Critics of the kill switch idea argue it would be too slow to respond to AI threats, as it took months to detect rogue agent attacks after they had largely concluded. For a kill switch to be effective, tech giants would need to be compelled to act at scale, particularly in the US, where the largest data centres are located. Kokotajilo acknowledged that kill switches were "better than nothing" but urged UK and global politicians to focus on diplomacy with the US to encourage meaningful regulation of AI development. A similar kill switch act is also being debated in the US, but President Trump has dismissed concerns about AI's threat to humans, calling the industry a "golden goose." The Cabinet Office stated it is continuing to take a "long-term, science-led approach" to understand and prepare for emerging risks from AI.