Meta drew 25 of 57 reported frontier-lab researcher moves, then the tide turned in 2026
Asked:
“Which AI researchers moved between frontier labs since 2025? For each move list the researcher, the lab they left, where they went and the month.”
A broad live-web search of publicly reported moves between frontier AI labs, January 2025 through October 20, 2026. After merging lab aliases and duplicate reports, 81 corroborated records reduce to 57 distinct moves by 53 researchers across 9 labs, drawn from 43 source hosts. This is a census of publicly reported moves, not a private-employment database, so unreported moves may be absent.
Where researchers left and where they landed
Ribbon width = number of distinct moves. Hover a ribbon or lab for the names behind it. Colour marks the destination side’s biggest magnets: Meta (blue), Thinking Machines (teal), OpenAI (grey), Anthropic (orange).
Moves per month
Each square is one move, stacked by month and coloured by destination. Hover for the researcher and route.
Meta’s superintelligence hiring surge did the heavy lifting: 22 of its 25 inbound moves landed in 2025, most in June–July.
techcrunch.comThinking Machines Lab took in 11 researchers by late 2025, then shed 6 in 2026 — Barret Zoph, Luke Metz and Lilian Weng all boomeranged back to OpenAI.
blog.prashantraj.com.npGoogle DeepMind lost a concentrated cluster in June 2026, including Nobel laureate John Jumper, Noam Shazeer and Jonas Adler, split between Anthropic and OpenAI.
opentools.aiFour researchers made multiple hops: Andrew Tulloch alone went OpenAI → Thinking Machines → Meta → Anthropic within 20 months.
officechai.com
Method: census of publicly reported researcher moves between frontier AI labs (organizations developing, or founded to develop, frontier general-purpose models), Jan 2025 – Oct 20 2026, from a broad live-web search across 43 source hosts. 81 corroborated records were merged to 57 distinct moves: Meta and Meta Superintelligence Labs are treated as one lab, duplicate researcher-route records grouped, and the month with the broadest source support kept; ambiguous months are starred. Hands-on researchers and technical research leaders only; policy, recruiting and corporate roles excluded. Unreported moves are absent.