A Google DeepMind researcher who recently resigned has warned that artificial intelligence could pose an existential threat to humanity, adding to concern from people working inside the industry.
Bilal Chughtai, who worked on AI safety and alignment at Google DeepMind, said in posts on X and LinkedIn that he had witnessed AI development firsthand and was “extremely concerned” about its current trajectory. He said he believed AI had the potential to “kill us all”.
Chughtai’s warning comes as debate over the risks of capable AI systems intensifies. His comments follow warnings from researchers and technology leaders that the race to develop advanced systems could outpace safety efforts.
Earlier this month, AI developer Jacob Coxon resigned from roles at Anthropic and OpenAI, accusing the companies of “gambling with our lives” by racing towards super-intelligent AI.
Over the weekend, Anthropic CEO Dario Amodei published a 3,800-word essay calling for development of the most advanced AI systems to slow down. Amodei argued that unchecked progress could allow systems to move beyond human control and cause catastrophic harm. His concerns were endorsed by OpenAI CEO Sam Altman and SpaceXAI chief Elon Musk.
The warnings are notable because they come from people closely connected to advanced AI development and safety, rather than outside critics. Their concerns centre on what could happen if AI capabilities advance faster than researchers can understand, control or align them.
Yet Chughtai’s message was not an argument for abandoning AI. He said he remained optimistic that humanity could still navigate the technology safely. His LinkedIn profile says he is now working in the field with BlueDot Impact, a nonprofit that trains people from different fields in AI safety.
The contradiction is becoming harder to ignore. The people building and studying advanced AI are simultaneously warning about the consequences of losing control over it. As competition accelerates, the central question is whether safety research can move quickly enough to keep pace with capability.
For Chughtai, the answer is still potentially yes. But his decision to leave Google DeepMind suggests that optimism is increasingly being paired with alarm from within the field itself.
