Meta has once again secured the services of two prominent OpenAI researchers, Jason Wei and Hyung Won Chung.
Both scientists, who joined OpenAI barely two years ago from Google Brain, have reportedly accepted roles in Meta’s newly formed Superintelligence Lab.
Jason Wei, whose work at OpenAI focused on reinforcement learning and model training techniques used in the o3 and Deep Research models, has already deactivated his internal OpenAI account, sources familiar with the matter confirmed to WIRED.
Hyung Won Chung, known for his expertise in reasoning models and agent-based systems, is making the same transition. Both researchers previously collaborated closely at Google before joining OpenAI simultaneously in 2023.
Meta’s aggressive recruitment strategy has become impossible to ignore. Offers reportedly reach up to $300 million per researcher spread across four years, comprising salaries, stock incentives, and bonuses.
CEO Mark Zuckerberg’s internal communication, revealed last month, openly acknowledged the company’s renewed focus on artificial general intelligence (AGI), with the Superintelligence Lab expected to lead this charge.
The lab itself, launched in June 2025, is co-headed by Alexandr Wang and Nat Friedman, two heavyweights formerly leading Scale AI and GitHub, respectively.
Their mission is to push Meta into the AGI race by building systems capable of human-level thinking, learning, and decision-making. Over the past month, Meta has pulled in at least 11 senior researchers from OpenAI, Apple, Anthropic, and Google DeepMind alone.
In Wei’s own words, shared in a recent social media reflection, “In life, (and when building AI models), imitation is good and you have to do it at first. But beating the teacher requires walking your own path and taking risks and rewards from the environment.”
That philosophy, it seems, now aligns with Meta’s approach.
Wei’s passion for reinforcement learning, where AI models learn through trial, error, and feedback, had earned him a reputation as a “diehard” advocate of the field during his stint at OpenAI. His departure, alongside Chung who worked on OpenAI’s o1 model and focused on agent-based reasoning, is a significant loss for Sam Altman’s team.
Not surprisingly, OpenAI isn’t sitting idle. After what CEO Altman described as Meta’s “distasteful” approach to recruitment, he confirmed Meta had indeed made nine-figure offers to his researchers, and the company responded with its own $500 million retention plan.
This includes accelerated stock options, sabbaticals, and across-the-board raises in an effort to keep its remaining experts grounded.
Multiple insiders describe Meta’s hiring spree as a deliberate strategy. Rather than hiring isolated individuals, the company has targeted researchers with close collaborative ties, Wei and Chung are the latest example, following a trio previously recruited from OpenAI’s Switzerland office.
Industry observers say intellectual capital now outweighs intellectual property. In the AI arms space, researchers are treated not unlike star athletes, “franchise players” whose departure can fundamentally shift the balance of innovation.
While Meta keeps building, OpenAI, Anthropic, and Google are working to lock down their talent pipelines. With salaries and incentives escalating, the competition for brains, not just algorithms, is changing the artificial intelligence development field.
