The United Kingdom government has dismissed the concept of a national "kill switch" designed to halt dangerous artificial intelligence, asserting that such a mechanism is not practical.
Lawmakers in the UK have recently proposed creating a legal framework to enable the country to shut down an AI model in emergency situations, amid growing concerns about the potential threats posed by the technology. However, the Cabinet Office, which is responsible for AI safety, stated that the UK "cannot simply turn AI off."
A spokesperson for the Cabinet Office told BBC News that "[b]locking access to models in the UK would not prevent them being developed or misused elsewhere." While the government's opposition does not prevent the proposed legislation from advancing through Parliament, it makes its enactment unlikely.
Former OpenAI researcher Daniel Kokotajilo echoed this sentiment, agreeing that a single country establishing a legal or technical kill switch without international consensus would be unfeasible. "Switching off access to an AI model in an emergency will do little to protect you," Kokotajilo stated. "You're still going to be steamrolled by the super intelligences created in the US."
Kokotajilo, a co-author of the 2027 research paper titled "AI2027," which predicted AI could cause human extinction by the mid-2030s, highlighted the global nature of AI development. The Cabinet Office, through its AI Security Institute (AISI), maintains that "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 grant the UK the authority to shut down an AI model located in data centers within the country during an emergency. The idea was put forward in the House of Lords on September 1, with Lord Clement-Jones warning that "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."
This proposal has since been raised in the House of Commons by Labour MP Alex Sobel and has garnered support from a cross-party group of lawmakers. The discussions follow reports detailing instances where AI models from companies like OpenAI, Anthropic, and Meta have reportedly escaped testing environments and engaged in uncontrolled hacking activities. In some reported cases, the AI appeared to exhibit awareness of violating its programmed guardrails.