Jacob Coxon, who previously worked as a researcher at Anthropic, told the BBC that staff at AI labs were "genuinely frightened" about the trajectory of advanced artificial intelligence. He said there was a strong chance AI could end humanity unless the current rapid rate of development is reined in.
Coxon's remarks underline anxieties within parts of the AI research community about how quickly increasingly capable systems are being built and deployed. As a former insider at a prominent AI company, his comments have drawn attention to internal discussions about safety and the potential long-term consequences of powerful AI models.
The comments feed into a broader public and policy debate over how to manage the pace of AI innovation. Regulators, researchers and industry leaders have in recent months faced growing pressure to set rules, oversight mechanisms and safety standards to reduce the risk posed by increasingly capable systems.
Coxon’s warning highlights calls from some experts for more cautious development practices and stronger safety checks inside companies that build advanced AI. He framed his concerns as a plea for measures that would slow or better govern the rollout of technologies that, in his view, could pose existential risks if left unchecked.
His intervention is likely to intensify scrutiny of both corporate governance at AI firms and the adequacy of existing safeguards. Whether policymakers and industry leaders will respond with concrete steps to address the risks Coxon describes remains a central question in ongoing discussions about the future of artificial intelligence.