Bilal Chughtai says advanced systems could pose an existential risk and advocates slowing down the race among the leading labs.
Bilal Chughtai, a researcher who worked on artificial intelligence safety at Google DeepMind, made his departure from the company public on Monday (14) and said he is “extremely concerned” about the technology's trajectory. He said he believes advanced AI systems have the potential to “kill us all”.
Chughtai worked on safety and alignment of artificial general intelligence (AGI), an area dedicated to ensuring that increasingly capable systems remain under control and follow objectives compatible with human interests. He left DeepMind in July 2026.
According to the researcher, models' capabilities are advancing faster than the techniques used to understand and control their behavior. Chughtai said that since he started working with AI in 2022, agents have begun to operate much more autonomously.
He argues that leading labs should slow the pace of development of the most advanced systems to allow safety mechanisms to keep up with the technology's evolution.
Warnings grow inside the labs themselves
The statement comes amid a string of warnings from researchers linked to the leading AI companies. Recently, professionals at Anthropic also called for greater caution over the possibility of future systems surpassing human capabilities in several areas.
After leaving DeepMind, Chughtai began working at BlueDot Impact, a nonprofit organization focused on training professionals in artificial intelligence safety.
