Bilal Chughtai
Bilal Chughtai is an AI safety researcher who worked as a research engineer at Google DeepMind on AGI safety and mechanistic interpretability until his resignation in July 2026. He is best known for his research on detecting strategic deception in AI models using linear probes and for publicly warning upon his departure that AI has the potential to cause existential harm to humanity. He is currently working at BlueDot Impact, a nonprofit focused on training people in AI safety.
“Google DeepMind researcher Bilal Chughtai, who focused on AI safety and alignment, has resigned, warning that the technology's 'default trajectory' could ultimately 'kill us all.'”
Source→“A leading AI safety researcher believes the field's current trajectory is existentially dangerous enough to leave a top lab entirely”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.