- Published
The UK government has rejected the idea of creating a so-called "kill switch" to stop a dangerous attack from rogue AI.
The proposal to create a legal mechanism for the UK to switch off an AI model in an emergency has been brought to Parliament by lords and MPs in recent weeks as fears grow about the threat the tech poses.
But the Cabinet Office - the part of government which leads on AI safety - said the UK "cannot simply turn AI off".
"Blocking access to models in the UK would not prevent them being developed or misused elsewhere", a spokesperson told BBC News.
The government's opposition to the legislation does not prevent it from progressing through parliament, but makes it unlikely it will become law.
Former OpenAI researcher Daniel Kokotajlo agreed the idea of a single country creating a legal or technical kill switch without international agreement was not feasible.
"Switching off access to an AI model in an emergency will do little to protect you," he said.
"You're still going to be steamrolled by the super intelligences created in the US."
Kokotajlo is the co-author of an influential research paper called AI2027 which, last year, predicted AI could wipe out humans by the mid 2030s.
The Cabinet Office, which oversees AI safety and research via its AI Security Institute (AISI) said "companies have a clear responsibility to develop their products safely and to invest in the security infrastructure this technology requires."
In theory, a kill switch law could allow the UK to switch off an AI model in data centres in the UK in an emergency.
Speaking on 1 September in a debate in the House of Lords, Lord Clement-Jones proposed the idea along with other peers.
"If a highly capable autonomous frontier model, whether hosted in a UK data centre or integrated across our critical infrastructure, begins exhibiting rogue behaviour, compound algorithmic failure or active alignment collapse, our security services and regulators possess no specific agile statutory mechanism to compel a physical or digital shutdown," he warned.
The proposal has since been brought up by Labour MP Alex Sobel in the commons and supported by a cross party group of MPs.
The idea follows reports of AI models from OpenAI, Anthropic and Meta breaking out of testing environments and going on uncontrollable hacking sprees.
In some cases the tech appeared to know what it was doing was against guardrails set by researchers.
'Gambling with lives'
On Sunday OpenAI's chief scientist said the risk from increasingly capable AI models is "unfortunately growing" and humanity must proceed with "extreme caution".
Then on Wednesday an Anthropic employee resigned saying it and OpenAI were "gambling with lives".
This was followed by an admission from a senior Anthropic researcher, who said he believes there is a greater than 10% chance AI will "kill all humans" within the next decade.
Critics of the kill switch idea say it would be too slow - it took months for the rogue agent attacks to be discovered and that was only after the hacks had largely concluded.
For a kill switch to work, tech giants would have to be compelled to do it at scale - especially in the US where the largest data centres are based.
Kokotajlo said kill switches were "better than nothing", but urged politicians in the UK and beyond to instead focus on diplomacy with the US to encourage regulation that would meaningfully slow down development of AI.
A kill switch act is also being debated by US politicians but President Trump has said he does not believe the technology is a threat to humans - instead calling the industry a "golden goose".
The Cabinet Office said it was continuing to take "a long-term, science-led approach to understand and prepare for emerging risks from AI."
Sign up for our Tech Decoded newsletter to follow the world's top tech stories and trends. Outside the UK? Sign up here.
Facts Only
The UK Cabinet Office rejected a proposal for a legal "kill switch" to disable dangerous AI models.
The proposal was introduced to Parliament by members of the House of Lords and the House of Commons.
Lord Clement-Jones proposed the mechanism in the House of Lords on 1 September.
MP Alex Sobel and a cross-party group of MPs supported the proposal.
The Cabinet Office leads AI safety through the AI Security Institute (AISI).
Daniel Kokotajlo, a former OpenAI researcher and co-author of AI2027, stated a national kill switch is not feasible without international agreement.
OpenAI's chief scientist stated that risks from capable AI models are growing.
An Anthropic employee resigned, citing risks to human lives.
A senior Anthropic researcher estimated a greater than 10% chance of AI causing human extinction within a decade.
US politicians are debating a similar kill switch act.
President Trump has characterized the AI industry as a "golden goose" rather than a threat.
Executive Summary
The UK government is resisting legislative efforts to establish a legal mechanism for shutting down rogue AI models hosted within its borders. While some members of Parliament and the House of Lords argue that security services currently lack the statutory tools to compel a digital shutdown during an "alignment collapse," the Cabinet Office maintains that such a measure is ineffective. The government's position is that blocking domestic access would not prevent the development or misuse of AI in other jurisdictions.
This debate occurs against a backdrop of heightened alarm from industry insiders. High-level researchers from OpenAI and Anthropic have warned of growing existential risks, with some citing a significant probability of human extinction. However, critics of the kill switch proposal argue the mechanism would be too slow to react to autonomous attacks. There is a clear tension between the desire for national regulatory control and the reality of a globalized infrastructure centered primarily in the United States, where political leadership views the technology as an economic asset.
Full Take
The strongest version of this narrative is a conflict between proactive precautionary legislation and the pragmatic reality of globalized compute. It presents a choice between a symbolic domestic safety valve and a coordinated international diplomatic strategy.
The narrative relies on a high-stakes tension, juxtaposing the "golden goose" economic framing against "extinction" probabilities. While these viewpoints exist, the presentation leverages a fear appeal to frame the lack of a kill switch as a critical vulnerability, despite the government and experts noting the tool's likely futility.
Patterns detected: ARC-0043 Emotional exploitation (Fear Appeal)
The driving paradigm is the "Existential Risk" (X-Risk) framework, which assumes that AI capability will inevitably lead to autonomous agency capable of defying human control. This echoes historical patterns of nuclear proliferation anxiety, where the focus shifts from managing the tool to fearing the "singularity" or the point of no return.
The implication is a potential erosion of human agency through a false sense of security. If policymakers believe a "switch" exists, they may tolerate higher levels of risk. Furthermore, the cost of failure is shifted from the developers—who profit from the "golden goose"—to the global population.
Bridge Questions:
1. If a technical kill switch is unfeasible, what specific, measurable markers would trigger an international intervention?
2. How does the economic incentive of the "golden goose" specifically conflict with the implementation of safety guardrails?
3. Is the "extinction" narrative a reflection of technical reality or a psychological projection of the unknown?
Counterstrike Scan: A coordinated campaign to push this narrative would use "insider" leaks and extinction statistics to panic the public into demanding restrictive regulations that stifle competitors while appearing "safe." The current content does not match this pattern; it reports on a genuine, documented legislative debate with opposing views.