Two AI safety researchers have left Google DeepMind amid concerns that highly capable AI systems could threaten humanity. Josh Engels will join the independent research group METR to focus on alignment for large models, while Bilal Chughtai is moving to the nonprofit BlueDot Impact, which provides AI safety education courses.

Engels spent about a year at DeepMind and posted on X that there is a high likelihood of AI causing major harm within the next five years. He said he will work on large-model alignment at METR.

Chughtai spent 18 months at DeepMind and posted on X that AI carries a risk of destroying humanity. He has joined BlueDot Impact, which offers AI safety education programs.

The departures come as technology leaders face renewed calls to slow AI development and tighten regulation. Sina Finance notes these exits follow other notable departures from leading AI labs, including Jacob Coxon and Joe Benton leaving Anthropic.