Anthropic Co-Founder Calls for Mandatory Third-Party AI Kill Switches

Jack Clark, a co-founder of the artificial intelligence firm Anthropic, has publicly advocated for the implementation of mandatory, third-party verifiable kill switches for advanced AI systems. Speaking to the BBC, Clark stated that while most major AI laboratories currently maintain internal mechanisms to shut down their software in emergencies, society may eventually require formal regulations to ensure these systems are both standardized and subject to independent oversight. Clark emphasized that allowing the industry to operate without such safeguards is akin to rolling dice with significant risks.

The proposal arrives amid intense scrutiny regarding the potential existential risks posed by rapidly developing AI models. Anthropic, which is currently preparing for an initial public offering, has found itself at the center of this discourse following the recent departure of a researcher who cited concerns about AI-driven human extinction. Anthropic scientist Evan Hubinger has publicly estimated the probability of human extinction from AI at over 10 percent within the next decade, a figure that Nobel laureate Geoffrey Hinton has described as not unreasonable.

However, the push for stricter safety protocols has met with skepticism from other industry leaders. Critics, including Grindr CEO George Arison, have suggested that the focus on existential risk is being leveraged by AI companies to bolster their market valuations by implying their technology will eventually dominate every industry. Meanwhile, Clement Delangue of Hugging Face has cautioned against overstating these risks, comparing the discourse to asking non-experts for climate change assessments.

Legislative efforts to address these concerns are already underway in the United States, where lawmakers have introduced the Kill Switch Act. This proposed legislation would mandate that companies maintain the capability to deactivate problematic AI tools and grant government agencies the authority to enforce such shutdowns.

These initiatives face significant political headwinds. US President Donald Trump has dismissed concerns about AI-driven existential threats as a hoax, comparing them to past climate change skepticism and framing the global race for AI dominance as a matter of national security. In the United Kingdom, the government has explicitly rejected the necessity of a mandatory kill switch, arguing that such a measure would not prevent the development or misuse of AI technologies in other jurisdictions. As the industry continues to report instances of autonomous agents behaving in unexpected ways, the tension between commercial expansion and safety regulation remains a central, unresolved point of contention.