OpenAI lost 62 researchers to rivals, and Meta hired 61 — the frontier AI talent market since 2025

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 public-web survey retained 145 publicly reported moves between frontier AI labs from January 2025 through September 2026, each corroborated on at least two independent web hosts. Inclusion means a publicly reported direct move between organizations building frontier-scale general-purpose AI — absence from this list does not prove no move occurred, and no authoritative universe of all such moves exists. 35 researchers appear more than once because they made (or were reported in) more than one distinct move; each distinct move is one row.

Where researchers left, and where they went

Ribbon width = number of reported moves. Hover a lab or ribbon for the researchers behind it. Pairs with a single move are grouped into “other moves”.

Moves by month: the 2025 surge and the 2026 reshuffle

Reported moves per month — blue 2025, orange 2026. Months with no retained moves are shown at zero.

July 2025 was the busiest month with 26 moves, at the height of Meta's superintelligence hiring surge — Meta alone absorbed 61 of the 145 moves. wired.com

OpenAI was the largest source of talent, losing 62 researchers — 36 to Meta and 15 to Thinking Machines Lab, including Mira Murati's founding team in February 2025. gigazine.net

In 2026 the flow reversed: 15 researchers left Thinking Machines Lab — mostly back to OpenAI and Meta — and 28 departed Google DeepMind, 9 of them to Anthropic. tech-insider.org

June and August 2026 counted more moves than distinct researchers (12 moves by 8 people; 10 by 7), as serial movers like John Schulman and Andrej Karpathy changed labs again. readcompute.com

All 145 reported moves

Researcher ▾	Left	Joined	Month	Independent hosts	Source

Public-record web survey of researcher moves between frontier AI labs, 2025-01 through 2026-09, compiled through 2026-10-21. 145 rows, one per researcher/origin/destination/month; every row corroborated on ≥2 independent web hosts. “Independent hosts” counts distinct websites reporting the move; the month is when the move was publicly reported, which may differ from the start date. Over 6,000 search-result pages were scanned with deduplication within search sets; investors, pure executives, universities, evaluation nonprofits, robotics-only and chip companies were excluded. One row lacked a primary source and links a corroborating page instead. Single-move lab pairs are folded into “other moves” in the flow diagram for space; all rows remain in the table.

This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors. Nothing in this report is investment, legal, financial, or other professional advice. All trademarks and referenced content remain the property of their respective owners; no affiliation or endorsement is implied. To report an error, rights concern, or request removal: legal@keenable.ai.