OpenAI is in damage control mode after losing a string of top researchers to rival Meta, in what is fast becoming one of the fiercest talent wars in the artificial intelligence industry.
Over the past week, Meta has hired at least eight researchers from OpenAI—including some deeply involved in its foundational AI reasoning models—prompting an unusually emotional internal response from OpenAI leadership and triggering a comprehensive review of compensation and retention strategies.
“Like Someone Broke Into Our Home”: Internal Memo Reveals Alarm
In a Slack memo obtained by Wired, OpenAI’s Chief Research Officer Mark Chen did not mince words describing what the situation feels like.
“I feel a visceral feeling right now, as if someone has broken into our home and stolen something,” Chen wrote.
The memo acknowledged the wave of high-profile exits and reassured employees that the leadership team—including CEO Sam Altman—has been working “around the clock” to stop the bleed. Chen said they were “recalibrating compensation,” and exploring “creative ways to recognize and reward top talent.”
The response came after reports that Meta had hired eight researchers from OpenAI in recent weeks, with recruitment efforts reportedly involving personal outreach from Mark Zuckerberg and offers reaching as high as $100 million—though Meta executives later said the figure had been exaggerated and mischaracterized as a simple signing bonus.
The departures are substantial in both number and profile. They include:
- Trapit Bansal, a central figure in OpenAI’s reinforcement learning and AI reasoning efforts
- Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai, known for their work in computer vision and multimodal AI
- Four additional researchers—Shengjia Zhao, Jiahui Yu, Shuchao Bi, and Hongyu Ren—whose expertise spans foundational model training and alignment
According to The Information and WSJ, the new hires are now part of Meta’s AI Superintelligence division, an ambitious unit set up to build general-purpose reasoning models that can rival OpenAI’s own GPT line and Google’s DeepMind efforts.
Meta has made no secret of its desire to dominate the next wave of AI innovation. Since launching its Llama 4 model in April—which reportedly fell short of internal expectations—Zuckerberg has doubled down, aggressively poaching talent and even attempting to acquire startups like Safe Superintelligence, Thinking Machines, and Perplexity AI.
Meta’s goal is to match and potentially surpass OpenAI and Google by developing cutting-edge reasoning models that can power enterprise tools, consumer products, and AI agents. The newly assembled team also includes former Scale AI CEO Alexandr Wang and former GitHub CEO Nat Friedman. Zuckerberg is said to be personally involved in some of these recruitment conversations, reportedly offering multimillion-dollar packages that include equity, long-term performance incentives, and high autonomy.
OpenAI’s Counteroffensive: Pay, Purpose, and Retention
In the face of these high-profile losses, OpenAI is revamping its internal compensation structure. Mark Chen’s memo acknowledged that staff felt demoralized, especially during the company’s designated “recharge week.” Still, he urged employees not to let “out-of-band offers” pressure their decisions, promising fairness and transparency in forthcoming changes.
The company’s leadership team is now engaging with employees who have received external offers, while also reviewing how to retain those who remain.
Chen said: “We’re recalibrating comp, and we’re scoping out creative ways to recognize and reward top talent… But I won’t do so at the price of fairness to others.”
A Deeper Power Struggle in the AI World
This talent war is not simply about salaries—it reflects a deeper rivalry between companies vying for supremacy in what is now the world’s most important and competitive technology race.
OpenAI, known for pioneering generative AI with ChatGPT, is under pressure to maintain its lead, especially as Google, Anthropic, and now Meta all move aggressively into the same space. Meta’s hiring spree comes at a time when AI expertise is in short supply and those who can build, align, and reason with frontier models are commanding extraordinary influence—and pay.
Meanwhile, Sam Altman has accused Meta of exploiting the open-source ethos while quietly attempting to gut rival teams. In a recent podcast, he lamented: “They’re trying to poach our people with crazy offers… but none of our best people have taken them up.”
However, with eight confirmed departures and morale shaken, that claim is now under scrutiny.