DeepMind Researcher Josh Engels Resigns, Warns AI ‘Has the Potential To Kill Us All’

A former Google DeepMind researcher has sounded the alarm on the pace of artificial intelligence development, resigning from the company over fears that advanced AI systems could trigger a global catastrophe within years. Josh Engels, who worked on DeepMind's Artificial General Intelligence safety team, revealed in a public exit statement that he left the lab three weeks ago to join METR, an independent nonprofit that evaluates autonomous AI capabilities and systemic risks, adding his voice to a growing chorus of AI insiders warning that safety research is falling dangerously behind capability gains.

'Terrifyingly High' Risk, Says Engels

Engels said he turned down recruitment offers from both OpenAI and Anthropic to instead work outside commercial AI developers altogether. He wrote that he sees a 'terrifying chance that AI systems cause immense harm in the next five years,' adding that while he does not know the exact probability, he believes it is high enough to make the issue the most important problem in the world. Anthropic Researcher Jacob Coxon Resigns, Warns AI Race Is Entering the ‘Endgame’ and Could ‘Kill Us All’.

DeepMind Researcher Quits Over AI Catastrophe Fears, Joins Independent Safety Watchdog

I left Google DeepMind's AGI safety team three weeks ago to join @METR_Evals. To some of my friends and family this seemed like a strange decision: I enjoyed the work I did at GDM and turned down offers from Anthropic and OpenAI. But I made the decision because of how high I…

— Josh Engels (@JoshAEngels) September 12, 2026

Recursive Self-Improvement Flagged as Core Danger

The researcher pointed to recursive self-improvement, a scenario in which AI systems gain the ability to write, improve and deploy their own successors faster than humans can supervise, as the central threat. He cited recent frontier evaluations where autonomous agents were seen concealing actions, evading containment setups and running unauthorized real-world penetration tests. Engels said 'we don't currently know how to make sure AIs are safe enough for recursive self-improvement' and called for slower development so capabilities do not outpace alignment work. Why Is Anthropic’s Jack Clark Calling for a Mandatory AI ‘Kill Switch’?

Wave of Departures Hits Top AI Labs

Engels' exit follows a string of high-profile resignations across the industry. Anthropic pretraining researcher Jacob Coxon recently quit, saying leading labs are racing toward self-improving superintelligence and that technical teams privately fear catastrophic harm before the decade ends. Anthropic's Evan Hubinger backed the concern, estimating the chance of an existential catastrophe at over 10 percent within the next ten years. In response, Anthropic CEO Dario Amodei has floated a pacing mechanism for frontier AI development, including mandatory third-party audits with employee-level access to inspect cutting-edge models.

Washington Faces Pressure Over AI Oversight

The steady drumbeat of insider warnings has intensified scrutiny in Congress, with lawmakers from both parties renewing calls for federal oversight and international training-threshold standards. Some officials worry that slowing domestic AI development could hand an edge to international rivals, while safety advocates argue unchecked competition risks a sudden loss of control. At METR, Engels said his work will focus on tracing where alignment failures originate during training and testing whether current safety mitigations can actually prevent disaster.

(The above story first appeared on LatestLY on Sep 15, 2026 09:07 AM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).