Director of frontier safety and governance at Google DeepMind, and previously the founder of the Centre for the Governance of AI, one of the earliest research groups dedicated to AI governance as a distinct field.
Work
Dafoe's research programme has centred on the political economy of transformative AI: how competition between states and firms shapes what gets built, and what institutional arrangements could alter those incentives. That framing distinguishes it from technical-alignment work, which asks whether a system does what its developers intend, and from applied-policy work, which asks what a given jurisdiction should require.
He is a co-author of DeepMind's From AGI to ASI (June 2026). The report's treatment of deliberate slowdown as one of six potential bottlenecks reflects the governance lens: it lists as counteracting factors that "economic and political pressures, and international race-dynamics may override slowdown pressures, particularly in light of lacking global coordination and effective global oversight and enforcement." That is a governance-side judgment appearing in a technical report, and it is notable for the direction it runs — the factor offered against a deliberate slowdown is the failure of coordination.
Relationships
- related: Google DeepMind — director of frontier safety and governance
- related: From AGI to ASI (Google DeepMind, June 2026) — co-author
- related: Frontier AI Governance, AI Governance (umbrella), US-China AI Competition: Different Races, Different Metrics