Meta hires top OpenAI researcher Trapit Bansal for its AI superintelligence lab, sources tell TechCrunch. Bansal left OpenAI in June, confirmed by an OpenAI spokesperson.
Bansal helped launch OpenAI’s reinforcement learning efforts and built their first AI reasoning model, o1. His arrival could boost Meta’s AI reasoning tech, which currently has no public models.
Meta’s AI superintelligence team also features ex-Scale AI CEO Alexandr Wang and could add former GitHub CEO Nat Friedman and Safe Superintelligence co-founder Daniel Gross. The goal: develop cutting-edge AI reasoning models rivaling OpenAI’s o3 and DeepSeek’s R1. 
Mark Zuckerberg has been aggressively hiring AI talent, reportedly offering $100 million packages. It’s unclear what deal Bansal accepted.
Three other ex-OpenAI researchers — Lucas Beyer, Alexander Kolesnikov, Xiaohua Zhai — recently joined Meta’s AI team, per The Wall Street Journal. Bloomberg adds former Google DeepMind’s Jack Rae and Sesame’s Johan Schalkwyk to the roster.
Zuckerberg also tried to buy AI startups with heavy research labs like Safe Superintelligence and Thinking Machines Labs but talks stalled.
OpenAI CEO Sam Altman said on a podcast Meta has been poaching OpenAI’s top talent but claimed:
"none of our best people have decided to take him up on that."
Meta has declined to comment.
AI reasoning models are vital for Meta’s ambitions to build smart AI agents across the company, competing with DeepMind and OpenAI. The competition heats up as OpenAI prepares to release an open AI reasoning model soon.
Meta’s hiring spree signals serious intent to push past rivals in AI research. The race just got faster.
