The race to build increasingly powerful artificial intelligence is facing a new warning from inside the industry. Another AI safety researcher has left a major technology company after raising concerns about the potential dangers of advanced AI. The latest departure comes amid growing debate over whether AI companies are moving faster than safety research can keep up.
The researcher has warned that highly capable AI systems could cause immense harm in the coming years if scientists fail to solve the problem of keeping them aligned with human goals.
The development has once again put the spotlight on a difficult question: Are AI companies developing increasingly powerful systems faster than researchers can make them safe? One name now at the centre of that debate is Josh Engels.
Who Is Josh Engels?
Josh Engels is an AI safety researcher who worked on the AGI safety team at Google DeepMind. He recently revealed that he left the company around three weeks ago. Engels said he enjoyed his work at DeepMind. However, he decided to leave because of what he described as the extremely high stakes surrounding advanced AI development.
In a post on X, Engels made a stark prediction. “I now think that there’s a terrifying chance that AI systems cause immense harm in the next five years,”
He said he did not know the exact probability. But he believed the risk was serious enough to make AI safety “the most important problem in the world”.
Why Did Josh Engels Quit Google DeepMind?
Engels said his concerns are linked to the race to develop AI systems that could eventually become more capable than humans across a wide range of tasks. He also revealed that he had rejected job offers from OpenAI and Anthropic. His decision suggests that his concerns extend beyond one particular AI company. Engels believes the broader race to develop increasingly powerful systems needs to slow down so safety research can catch up.
Josh Engels Warns About AI Self-Improvement
A major concern for Engels is recursive self-improvement, or RSI. This is the idea that AI systems could help develop even more advanced AI systems. Those systems could then improve themselves further, creating a rapid feedback loop.
“The AI companies are all trying to build superintelligence: systems vastly better than humans at everything,” Engels wrote.
He acknowledged that such technology could be extremely useful for humanity. But he warned that researchers still do not know how to guarantee that these systems will remain safe.
“A misaligned RSI loop could be catastrophic,” he added.
Where Is Josh Engels Working Now?
After leaving Google DeepMind, Engels joined METR, a nonprofit organisation that evaluates advanced AI systems and studies their potential risks. He plans to research AI misalignment and examine whether current safety methods are strong enough. He also wants to understand whether researchers are making enough progress to solve the AI alignment problem.
His move comes shortly after other AI safety researchers raised similar concerns, adding fresh pressure on the industry to reconsider the pace of the AI race. For Engels, the biggest concern is not simply how powerful AI can become. It is whether humans will be able to control that power once it becomes significantly more capable.
Sudeep Singh Rawat is a Copy Editor at NewsX Digital with over five years of experience in digital journalism. He covers national and international news, with a focus on politics, crime, technology, public affairs, and trending topics. Before joining NewsX Digital, he worked with Business Standard, The Sunday Guardian, and other digital media organisations. His reporting focuses on breaking news, enterprise stories, and explainers that inform readers through accurate, balanced, and context-driven journalism.
