creators

Rohin Shah: AGI safety lead at Google DeepMind

Rohin Shah leads the AGI Safety and Alignment team at Google DeepMind. He did his PhD at CHAI in Berkeley and wrote the 173-issue Alignment Newsletter.

Rohin Shah leads the AGI Safety and Alignment team at Google DeepMind, the group that works on reducing existential risks from AI systems; a July 2026 hiring post he co-signed describes it as 'the team at GDM, led by Rohin Shah'. He has worked on AI safety since 2017 and did his PhD at the Center for Human-Compatible AI at UC Berkeley, on AI systems that can learn to assist a person who has not spelled out what they want. His 2019 paper 'Preferences Implicit in the State of the World', with Dmitrii Krasheninnikov, Jordan Alexander, Pieter Abbeel and Anca Dragan, argued that an environment humans already live in encodes their preferences. He wrote the Alignment Newsletter, which summarised and commented on new alignment research for more than 2,600 subscribers across 173 issues between April 2018 and July 2022 and is now on indefinite hiatus. He is involved in effective altruism, having been co-president of EA UC Berkeley in 2015-16 and run EA UW in 2016-17, and he described the work of running AGI safety at Google DeepMind on the 80,000 Hours podcast, in an interview recorded in December 2025 and released in June 2026.

Specialization

AI alignment, AI safety, reward learning, AGI risk

Country

United Kingdom