Technology and Science13 September, 20263 min read
Jack Clark Urges Mandatory AI Kill Switch With Independent Verification, BBC Says
This article is an automated aggregation of coverage from news outlets. Our editorials offer in-depth analysis curated by the NewsCord team.
Jack Clark calls for mandatory AI kill switch with independent third-party verification. UK government rules out AI kill switch amid safety concerns and talks.
Kill switch debate heats up
Jack Clark, a Brighton-born tech founder and co-founder of US AI company Anthropic, told the BBC that AI firms should be required to have an emergency "kill switch" with independent third-party verification. Clark said Anthropic already has a kill switch and warned, "The thing about AI is it moves much, much faster than any normal technology," as he called for "urgent action now." The debate over safety and pace intensified after Anthropic CEO Dario Amodei called for the pace of development to slow down, with the concerns echoed by Sam Altman, CEO of OpenAI, and Elon Musk, founder of xAI.
“Artificial intelligence companies should be required to have an emergency 'kill switch'.”
In response, US President Donald Trump dismissed the warnings, labelling them a "sick conspiracy" designed to keep China ahead of the USA in technological development. China rejected AI slowdown calls as Trump dismissed guardrails, with Chen Yixin, head of China's Ministry of State Security, warning in an article published by the state-run magazine China Cyberspace that AI could threaten the Chinese Communist Party's grip on power.
UK rejects shutdown mechanism
The UK government rejected calls for an AI "kill switch" that would allow ministers to shut down AI systems during an emergency, with the Cabinet Office telling the BBC it "cannot simply turn AI off." The Cabinet Office said preventing access to a model or shutting down certain infrastructure would not stop the same model being developed and hosted elsewhere, and the government implied responsibility lies with AI companies to build products with safeguards. Former OpenAI researcher Danial Kokotajlo argued that switching off access to an AI model in an emergency would do little to protect users, saying, "You're still going to be steamrolled by the super intelligences created in the US."
In the same broader policy debate, Liberal Democrat peer Lord Clement-Jones proposed an amendment to the Cyber Security and Resilience Bill for a statutory mechanism, but the government said ministers would continue taking a "long-term, science-led approach" to understanding emerging AI threats. Meanwhile, BBC coverage of Anthropic co-founder Jack Clark said a kill switch "might want to eventually pass rules around," and Clark asked whether lawmakers should mandate verifiable shutdown requirements for companies.
Who controls the risk
Anthropic CEO Dario Amodei told CBS News, "I won't lie to you – there are real dangers," and said the industry had lied "about the fact that this technology had risks." Amodei said the warning sign is that AI development has been on an exponential curve that is "starting to get steeper," adding, "we need to slow down" without shutting everything down. He argued Anthropic would give independent evaluators permanent "employee-like" access to models, describing it as "like a food inspector," while also supporting federal regulation and saying the "AI kill switch could be a good idea."
“Should you mandate for companies to definitely have a kill switch?”
In the BBC account, Jack Clark said most labs have different ways of being able to pull the plug, but lawmakers may need to enforce having one, asking, "Should you mandate for companies to definitely have a kill switch?" The stakes are framed as both human safety and geopolitical competition, with Newsweek reporting that Chen Yixin warned AI poses risks including foreign espionage and misinformation campaigns to cyberattacks on critical infrastructure, while Trump posted that "WHOEVER WINS AI, WINS!"