and the distribution of digital products.
Meta recruits top OpenAI talent to build new reasoning models
Meta hired Trapit Bansal, a former OpenAI researcher, to advance its AI reasoning models within its new AI superintelligence unit, confirmed by sources familiar with the matter and an OpenAI spokesperson.
Trapit Bansal joined OpenAI in 2022, playing a significant role in initiating the company’s work on reinforcement learning alongside co-founder Ilya Sutskever. He is also recognized as a foundational contributor to o1, OpenAI’s initial AI reasoning model. Bansal’s LinkedIn profile indicates his departure from OpenAI occurred in June 2025. His addition is anticipated to bolster Meta’s AI superintelligence lab, which includes figures such as Alexandr Wang, former CEO of Scale AI.
Meta is also reportedly seeking to incorporate Nat Friedman, former GitHub CEO, and Daniel Gross, co-founder of Safe Superintelligence, into this unit. The intention is for Bansal to aid Meta in developing a frontier AI reasoning model capable of competing with established industry leaders, including OpenAI’s o3 and DeepSeek’s R1. Currently, Meta does not offer a publicly accessible AI reasoning model.
Mark Zuckerberg has pursued an aggressive hiring strategy in recent months to expand Meta’s AI team. This strategy has involved offering compensation packages valued at $100 million to attract top researchers. The specific terms of Bansal’s compensation package were not disclosed. This recruitment drive has successfully secured notable AI research talent. In addition to Bansal, three other former OpenAI researchers, Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai, have joined Meta’s AI superintelligence team recently, as reported by The Wall Street Journal. Bloomberg also reported that Jack Rae, previously with Google DeepMind, and Johan Schalkwyk, formerly a machine learning leader at Sesame, have also joined Meta’s AI team.
Meta wins AI copyright fight with authors
Meta reportedly attempted to acquire several startups possessing advanced AI research capabilities to further staff its new AI unit. These included Safe Superintelligence, founded by Ilya Sutskever; Thinking Machines Labs, associated with Mira Murati; and Perplexity. However, discussions for these acquisitions did not advance to a definitive stage. Sam Altman, CEO of OpenAI, stated in a recent podcast that Meta has been attempting to recruit top talent from his company. He asserted that “none of our best people have decided to take him up on that.” A Meta spokesperson declined to comment on these specific claims.
AI reasoning models represent a crucial development area for Meta’s AI superintelligence team. Over the past year, OpenAI, Google, and DeepSeek have released high-performing AI reasoning models, demonstrating advancements in software capabilities. These models enhance AI performance in benchmarks and real-world applications by utilizing additional time and computing resources to process problems before generating solutions.
Meta’s AI superintelligence lab is poised to become a central internal group supporting various company products, similar to Google’s DeepMind unit. Meta is also engaged in an initiative to develop AI agents for business purposes under the leadership of Clara Shih, former Salesforce CEO of AI. The development of competitive agents necessitates robust, frontier AI reasoning models to power them. The integration of Bansal and other key AI researchers is intended to accelerate Meta’s progress in this competitive field. OpenAI’s impending release of an open AI reasoning model in the coming weeks may intensify competition for Meta’s open AI offerings.