UK government dismisses “kill switch” proposals for AI safety
Proposals to introduce a so-called “kill switch” to protect against an attack from rogue artificial intelligence have been turned down by Andy Burnham’s government, reports BritPanorama.
On Friday, the Cabinet Office voiced its resistance to establishing a new “last-resort power” that would permit the state to shut down an autonomous model during an emergency, stating the UK “cannot simply turn AI off”.
This decision comes amid escalating alarms over technological risks, including expert warnings that AI could pose existential threats, with some suggesting it could “wipe out humanity within the decade”.
Earlier in September, Liberal Democrat Lord Clement-Jones submitted an amendment to the Cyber Security and Resilience Bill creating a legal mechanism for state intervention should a model malfunction. He argued for “a necessary emergency kill switch” to ensure the Government has the authority to direct an emergency shutdown when lacking clear statutory power.
However, a government spokesperson responded by asserting: “Britain cannot simply turn AI off and blocking access to models in the UK would not prevent them being developed or misused elsewhere.” They emphasized the responsibility of companies to ensure the safe development of their products and invest in necessary security infrastructure.
In their statement, officials highlighted that the UK would continue to adopt a long-term, science-led approach to understanding emerging risks from AI, including through initiatives run by their AI Security Institute.
The spokesperson also noted that the Cyber Security and Resilience Bill aims to enhance UK cyber defenses and the cybersecurity of essential public and digital services, including data centres.
Nevertheless, officials indicated that the Government is examining whether further targeted interventions are necessary to address the most significant risks linked to AI technology. Any measures implemented would require careful consideration to ensure they are proportionate.
Earlier this week, Prime Minister Burnham recognized the “risks to national security” posed by AI, coinciding with alarming statements from Evan Hubinger, a top safety researcher at Anthropic. He mentioned the belief that AI could “kill all humans” within the decade.
Burnham has also elevated AI safety minister Kanishka Narayan’s role by providing him a position at the Cabinet table, asserting that this move has “lifted the level of the conversation about artificial intelligence within Government.”
Hubinger’s comments followed on social media, contending: “We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade.”
In a related context, his concerns were echoed by fellow Anthropic researcher Jacob Coxon, who left the firm lamenting that companies were racing towards self-improving superintelligence and not acting responsibly. He characterized this as “gambling with our lives.” He stressed that the danger posed by AI is underestimated and warned, “The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt.”
Further complicating the landscape, reports from the Financial Times have indicated that US tech giant Anthropic withheld its latest AI model, Claude Mythos 5.1, from the UK’s AI Safety Institute, providing access only to vetted American organizations.
Contact has been made with Anthropic and OpenAI for further comments regarding these developments.
