VP of Alignment at Anthropic. Former head of the Superalignment team at OpenAI, from which he resigned in May 2024 alongside Ilya Sutskever, citing insufficient prioritization of safety research.
Leike's public resignation from OpenAI was the single most damaging credibility blow to the lab's safety narrative — more consequential than any external criticism could ever be. He didn't just leave; he told the world exactly why, and the gap between OpenAI's stated 20%-of-compute safety commitment and the reality his team experienced became impossible to spin. His move to Anthropic was a signal trade: one lab's loss of alignment credibility became another's recruitment win. Watch what his team ships — it will define whether alignment research can actually keep pace with capabilities.
No recent LinkedIn activity tracked yet.