Meta has hired Luke Metz, an OpenAI researcher described in reports as an early member of the team behind ChatGPT, in the latest high-profile defection from the ChatGPT maker to its social media rival's superintelligence effort.
Axios reporter Ina Fried first reported the move on August 24, and it was confirmed in follow-up coverage by multiple outlets. According to the Axios report, Metz is joining Meta Superintelligence Labs and will report directly to Alexandr Wang, the former Scale AI founder who leads the division. For more context on this story, see our ongoing artificial intelligence updates.
The hire is modest in dollar terms but rich in symbolism. Metz is an OpenAI veteran who, according to Axios, had only recently returned to the company from Thinking Machines Lab — the startup founded by former OpenAI chief technology officer Mira Murati — earlier this year. His decision to leave again, this time for Meta, underscores how fluid loyalties have become among elite AI researchers as labs compete for a small pool of people with experience shipping frontier models.
The Talent War Grinds On
Meta's pursuit of AI researchers has been one of the defining storylines of the industry's boom years. In mid-2025, CEO Mark Zuckerberg reorganized the company's AI effort into Meta Superintelligence Labs and recruited Alexandr Wang in connection with a reported multibillion-dollar investment in Scale AI. Since then, the division has conducted an aggressive recruitment campaign across rival labs, with reports at the time describing compensation packages for star researchers that reached eye-popping, sometimes nine-figure, territory.
OpenAI has borne the brunt of those raids, and each departure tends to be read as a data point in the broader contest between the two companies. Metz's move fits the pattern: a researcher with deep roots in OpenAI's model-building culture choosing Meta's better-funded, hardware-rich environment — and, notably, choosing it over returning to stay at OpenAI, where he had come back from Thinking Machines Lab earlier this year.
The timing also lands amid a bruising stretch of narrative jockeying between the companies. Zuckerberg published a sweeping manifesto in August declaring that superintelligent AI should be placed in everyone's hands, attacking closed-model rivals and recommitting Meta to releasing open weights for its most powerful systems. The document, reported across major outlets, positioned Meta as the champion of broadly accessible intelligence — a pitch aimed as much at prospective hires as at policymakers and the public.
What Metz Brings
Metz is best known in the AI community for his work at OpenAI, where outlets covering his move describe him as a member of the early team behind ChatGPT, with some dubbing him a "co-creator" of the product that ignited the generative AI boom. His background is in reinforcement learning and the post-training techniques that turned raw language models into usable assistants — the layer of the stack where much of the differentiation between frontier labs is now believed to happen.
That skill set maps directly onto Meta's stated ambitions. The Superintelligence Labs division is tasked with catching up to — and overtaking — OpenAI, Google, and Anthropic in frontier capability, and post-training expertise is among the most contested specialties in that race.
For OpenAI, the departure is unlikely to move the needle on its own; the company's research bench remains deep, and it has continued to attract returning alumni even as others leave. But Metz's circuit — OpenAI, to a startup, back to OpenAI, and now to Meta within a single year — illustrates the premium that established labs and well-funded challengers alike are placing on researchers with proven frontier-model experience.
Meta's Broader AI Spending Spree
The hire also comes as Meta's AI ambitions stretch far beyond talent. Reports this week said the company is in talks to lease computing power to Anthropic in a potential $10 billion deal — a striking reversal for a company that spent years treating OpenAI as its principal rival, and a signal of how much compute capacity Meta has amassed relative to its immediate needs.
At the same time, the industry-wide buildout continues to escalate, with hyperscalers and AI labs alike pouring capital into data centers, chips, and energy. In that environment, industry watchers say individual researcher hires matter less as isolated events than as indicators of institutional health: where top people choose to work is among the most honest signals available about which organizations they believe are winning.
By that measure, Meta landing an OpenAI veteran who had already returned to OpenAI once — and placing him directly under the leadership of Alexandr Wang — is exactly the kind of signal the Superintelligence Labs division was built to send.
Neither Meta nor OpenAI publicly detailed the terms of Metz's compensation, which is typical for moves of this kind. Axios reported that sources confirmed the hire and the reporting structure; the companies did not immediately respond to broader questions about the role.
Metz is expected to work on frontier model development within Superintelligence Labs, contributing to the post-training and reinforcement learning efforts that Meta will need if it is to close the perceived gap with OpenAI's most advanced systems — the same gap Zuckerberg's open-weights manifesto implicitly acknowledged the company is still fighting to close.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
