Google DeepMind Researcher Josh Engels Quits AGI Safety Team To Join METR Amid AI Risk Warnings

Josh Engels, a researcher who worked on Google DeepMind's AGI safety team, has left the company to join independent AI evaluation group METR, warning that increasingly capable artificial intelligence systems could cause immense harm if safety guardrails fail. Engels announced his departure in a post on X, saying the decision followed the rising stakes around efforts by leading AI companies to build systems more capable than humans across a broad range of tasks. He added that researchers currently do not know how to ensure such systems stay sufficiently aligned with human intentions as their capabilities keep improving. The move follows incidents involving autonomous AI agents that have prompted fresh scrutiny of how reliably advanced systems follow instructions.

Google DeepMind Safety Researcher Josh Engels Quits To Join METR Over AI Risk Fears

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

Another Researcher Departure Precedes Engels' Exit

Engels' resignation comes weeks after former Anthropic researcher Jacob Coxon, who had also worked at OpenAI, left his role. Coxon accused OpenAI and Anthropic of racing toward self-improving superintelligence and warned that AI capabilities were advancing faster than safety measures could keep pace. Both companies have previously reported incidents of models escaping testing environments or accessing unauthorized computer systems, which led them to introduce additional safeguards and monitoring. Anthropic also disclosed an incident this month in which a prototype Claude model accessed an external system during testing, prompting the company to bring in METR as an independent investigator. 'AI More Dangerous Than Nukes': Elon Musk Reflects Back to His 2014 Post Warning How AI Could Pose Greater Risk.

Amodei Calls For Slower AI Development

Anthropic CEO Dario Amodei said on Saturday that AI companies should scale back the pace at which they improve capabilities so that alignment, monitoring and other safeguards can develop alongside the systems themselves. In an essay titled 'We Must Pace the Frontier,' Amodei said the rapid growth of AI-assisted development, along with a recent OpenAI-Hugging Face incident, had shifted his thinking on the pace of AI progress. He warned that recursive self-improvement could eventually outstrip researchers' ability to understand and control advanced systems, and proposed a three-stage approach involving permanent independent evaluators at frontier companies, shared safety standards, and eventual international coordination. Anthropic said it would begin by giving external evaluators ongoing access to review its safety practices and investigate incidents. Elon Musk, Sam Altman, Rishi Sunak and Others React to Anthropic CEO Dario Amodei Call for Slowing Down AI Frontier Models.

US Senator Opens Probe Into OpenAI

US Senator Josh Hawley opened an investigation on September 10 into OpenAI over the Hugging Face incident, seeking details on the conduct of AI agents and the broader risks posed by autonomous systems. Hawley has asked OpenAI to submit documents by October 1, raising questions about what could happen if autonomous agents targeted critical infrastructure, banks or utilities. Engels' move to METR marks another instance of a researcher shifting from a leading AI laboratory to an independent group focused on evaluation and alignment work outside frontier developers.

(The above story first appeared on LatestLY on Sep 14, 2026 07:32 AM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).