Every AI lab thinks it's the responsible one, and safety researcher Ryan Greenblatt says that's what keeps the arms race going

Redwood Research chief scientist Ryan Greenblatt estimates a 50 to 60 percent probability of an AI takeover if current development trajectories persist. He notes that individual AI labs often perceive their own safety protocols as superior, a mindset that facilitates a competitive race rather than collective caution. This highlights a disconnect between industry acceleration and the historical risk thresholds typically associated with existential threats.