One signal

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.