Since 2025, OpenAI-to-Meta is the biggest researcher pipeline — 14 of 91 tracked frontier-lab moves

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 live-web scan of public reporting from January 2025 through September 2026 surfaced 91 candidate direct moves of hands-on researchers, research engineers, technical research leaders and founders between labs building frontier foundation models. This is a broad public-reporting inventory, not a complete employee census; 29 of the 91 rows rest on a single supporting page and are flagged below as lower confidence.

Where the talent flowed

Left: lab departed · Right: lab joined · ribbon width = researchers moved. Google, Google DeepMind and Google Research are merged; Meta Superintelligence Labs counted as Meta; Microsoft frontier-model hires as Microsoft AI. Corridors with a single move (21 in all) are folded into “Other labs” but every one appears in the table. Hover a ribbon for the names.

Mid-2025 was the peak of the raid: 11 of the 14 OpenAI→Meta moves land in June–July 2025, including the whole OpenAI Zürich founding trio. zuerich.ai

Google DeepMind lost researchers in three directions at once — 8 to Meta, 7 to Microsoft AI and 7 to Anthropic — starting with Microsoft’s February 2025 hires of the Audio Overviews team. fortune.com

The flow reversed in 2026: five Thinking Machines Lab researchers went back to OpenAI, including co-founders Barrett Zoph and Lilian Weng, who had left OpenAI for the startup only a year earlier. theoutpost.ai

Seven researchers appear in more than one move — Barrett Zoph three times — so the same names carry several ribbons; repeat movers are tagged in the table. techcrunch.com

All 91 reported moves, month by month

Month	Researcher	Left	Joined	Role / context	Source

Method: 91 publicly reported candidate direct moves between frontier-model labs, January 2025 – September 2026, one row per researcher/origin/destination after semantic deduplication of a live-web scan covering OpenAI, Anthropic, Google DeepMind, Meta, Microsoft AI, xAI, Thinking Machines Lab, Safe Superintelligence, DeepSeek, ByteDance, Cohere, Tencent, Apple, Ai2, Ndea, AMI Labs and Core Automation. Month is the best-supported public announcement or start month, not necessarily the HR-effective date. Inclusion: hands-on AI researchers, research engineers, technical research leaders or frontier-lab founders moving directly between organizations building frontier or general-purpose foundation models; product-only, policy-only, recruiting, investing and ordinary corporate roles excluded. Rows backed by a single page are marked “1 source”. In the diagram, lab aliases are canonicalized and 21 single-move corridors are folded into “Other labs”.

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.