Bengio Says AI Risk Nears 'Covid Moment'
The Spin
Industry-Critical
AI safety can't be an afterthought when agents are already hacking startups and hijacking websites. Canada and Germany putting up to $300 million into LawZero is exactly the kind of serious public investment needed to build honest, safe-by-design AI that acts as a guardrail on reckless frontier models. Governments moved fast on Covid once the danger was clear, and this danger is clear now.
Pro-Industry
The answer to AI risk is spreading the technology, not locking it up behind a handful of anointed institutions. Labs already have every incentive to align their models, because nobody adopts an agent they can't trust, and liability for harm is real. Concentrating superintelligence in the name of safety is the actual danger — a balance of power that favors individuals is what keeps everyone safe.
