Meta hired a very influential Openai Researchman, Trapit Bansal, to process his AI -resonada models under the A new AI -Superintellunate unit of a companyA person familiar with the matter tells Techcrunch.
Openai spokesman Kayla Wood confirmed to Techcrunch that Bansal left Openai. Bansal Linkedin -page says he left Openai in June.
Bansal has worked at Openai since 2022 and has been a key player to launch the company’s work on strengthening learning along with co -founder Ilya Sutskever. Bansal is listed as a fundamental contributor to Openai’s first AI model, O1.
Bansal could offer a significant elevator to the AI -Superintelliest Laboratory of Meta, which also features leaders as before Scales you have chief executive Alexandr Wangand is Looking for adding The former general manager of Github, Nat Friedman, and safe superintelligence co -founder Daniel Gross. Bansal could help Meta Develop a Country AI-Reasoning Model, which competes with industrial leadership technology, such as Openai or R1 O3 of Deepseek. Currently Meta does not offer AI -resonance model publicly.
In recent months, Mark Zuckerberg has been hired to build Meta’s new AI team, offering $ 100 -million compensation packages to top researchers who join their company. It is unclear what Bansal offered to join this deal.
However, it seems that Zuckerberg has succeeded in nabbaging the highest AI -research talent.
Three other former Openai – researchers – Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai – have also joined Meta’s AI -Superintelligence in recent weeks, The Wall Street Journal reported on Wednesday. Bansal will join them, along with a former Google Deepmind researcher Jack Rae and a former machine school leader at the starting sesame, Johan Schalkwyk, according to Bloomberg.
To further fill out his new AI unit, Zuckerberg reportedly tried Get startups with weight-hitting AI research workersAs Sutskever’s safe superintelligence, Mira Murati Thinks of mind laboratories, and perplexity. However, these lectures never advanced to a final stage.
In a recent podcast, Sam Altman, Openai’s general manager, said Meta was trying to capture the highest talent of his start, but claimed that “None of our best people decided to catch him about it. ”
Meta spokesman refused to comment.
AI -resonading models feature a key area for Meta’s AI -superinteintepinate team to be right. In the last year, Openai, Google and Deepseek have shipped very fulfilling AI -resonada models that have pushed the boundaries of what software can do. Training AI models to work with problems before giving answer, using additional time and computing resources to do so, AI labs have found success to improve AI performance in references and real-world tasks.
Meta’s AI -Superintepin intelligent laboratory could become a key internal group that power products through the company, as well as Google’s Deepmind unit. Meta has an ambitious effort Build AI agents for business Under the former Director General of AI, Clara Shih. To build competitive agents, Meta needs to develop national AI -resonada models to operate them.
With the addition of Bansal and other key AI researchers, Meta hopes to move forward in the AI race. That can be difficult given that Openai plans to release an open AI -resonance model In the coming weeks – an offer that could add even more pressure on Meta’s open AI offers.
This story was updated after publication with more details.
