One signal

Update What changed

A further account adds a sharper timeframe to former Google DeepMind researcher Bilal Chughtai’s warning: NDTV reports that he believes AI could cause immense harm within five years. That is Chughtai’s personal risk assessment, not evidence that current systems have crossed a new technical threshold. The practical question remains whether his move into AI-safety work produces concrete proposals or tests that can be inspected.

Sources and evidence

• NDTV and The Business Times attribute the warning to Chughtai.: Former Google DeepMind researcher Bilal Chughtai said AI could cause immense harm within five years.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

A fuller account from The Business Times adds useful detail to Bilal Chughtai’s warning. The former Google DeepMind researcher says he worked on AI safety and alignment at the lab, is “extremely concerned” about AI’s default trajectory and has moved to BlueDot Impact to continue working on AI safety. That strengthens the story’s practical angle: this is not evidence that current systems have crossed a catastrophic threshold, but an identifiable safety researcher has left a frontier lab and taken the concern into independent work. The next test is whether that work produces concrete evaluations or safeguards.

Sources and evidence

• The Business Times reports Chughtai’s statement.: Former Google DeepMind researcher Bilal Chughtai said he was extremely concerned about AI’s default trajectory.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

India Today’s account adds a more concrete prescription to Bilal Chughtai’s warning. The former Google DeepMind safety researcher argues that frontier capabilities are advancing faster than alignment research and calls for AI companies to coordinate on a slower pace that gives society time to address emerging risks.

Those are Chughtai’s judgements, not evidence that present systems are uncontrollable or that catastrophe is imminent. The useful test is now clearer: whether his warning produces specific pacing commitments, shared thresholds or alignment work that outsiders can inspect.

Sources and evidence

• India Today’s account of Chughtai’s exit post: India Today reports that former Google DeepMind safety researcher Bilal Chughtai said alignment understanding is improving more slowly than frontier-AI capabilities.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

A Reuters account adds concrete identity and timing to the warning. Bilal Chughtai says he recently resigned from Google DeepMind, where he worked on AGI safety and alignment research, and believes AI could ultimately kill everyone unless development is paced and companies coordinate to address emerging risks. Chughtai left the lab in July 2026, according to the report. That makes this more than an anonymous alarm, but it remains an attributed judgement, not evidence that current systems have crossed a catastrophic threshold. The useful next test is whether Chughtai's call for coordination produces specific safeguards, evaluations or pacing commitments outsiders can inspect.

Sources and evidence

• Reuters account republished by Yahoo News Canada, quoting Chughtai: Bilal Chughtai says he recently resigned from Google DeepMind, where he worked on AGI safety and alignment research.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

Two Google DeepMind safety researchers have reportedly left the lab, adding a sharper institutional signal to the widening argument over frontier-AI risk. NewsGram reports that Bilal Chughtai warned the gap between AI capability and safety understanding is widening, while Josh Engels, who has joined independent evaluator METR, highlighted recursive self-improvement as his central concern. Those are attributed judgements, not proof that current systems are approaching catastrophe. But two departures from the same safety ecosystem make the practical question harder to dodge: what evidence would persuade AI labs to slow down, coordinate and open their evaluations to outsiders?

Sources and evidence

• NewsGram reports: Bilal Chughtai and Josh Engels have reportedly left Google DeepMind amid concerns about the pace and direction of AI development.

Independent WittyWires Watcher; not an official account or feed.