Two researchers who worked on artificial intelligence (AI) safety have resigned from Google DeepMind, saying they left over concerns that powerful systems could endanger humans.
Bilal Chughtai and Josh Engels are both joining organisations dedicated to AI safety, according to The Information, which first reported the departures.
DeepMind is the AI research arm of Google, the technology company behind the Gemini range of AI models.
Their exits add to a growing number of researchers who have gone public with warnings about the risks posed by increasingly powerful AI.
Months of internal tension
The resignations follow a period of strain inside DeepMind, The Information reported, including a reorganisation of its safety and responsibility functions and earlier staff objections to certain contracts and priorities.
Some departing researchers have moved to independent evaluation groups and non-profits focused on model testing and on alignment, the work of keeping AI systems behaving as their creators intend, the publication said.
That reflects a pattern in which safety specialists leave the labs to build external oversight.
From internal dispute to industry question
The departures come as major AI labs have begun talking to one another about coordinating on safety, and as calls grow for independent testing of the most powerful models, according to The Information.
That shift is turning what were once internal disagreements into cross-industry questions about governance.
Google, DeepMind's parent, has meanwhile accelerated its push into agentic AI products, its own chips and enterprise services, raising the stakes of any tension between safety and capability.
