Meta hires key OpenAI researcher to work on AI reasoning models

Meta, a highly influential Experai researcher, hurriedly recruiting to work on expensive, AI reasoning models Company’s new AI SuperTlay unitA person familiar with the issue explains to TechCrunch.

Openai Press Secretary Kayla Wood, Bansal’s Openai has confirmed Techcrunch. Dishonorable LinkedIn page He says Openai left in June.

BANSAL has worked in Openai since 2022, and the co-founder of the company was a major player in the study work in strengthening with Ilya Sutskever. It is listed in the first AI substantial model of Openai, as a foundation contributor in O1.

BANSAL, such as the past, can offer a significant lifting of Meta to the AI ​​Supervise Laboratory, which is also leaders There is a CEO of Trade Alexandr Wang and Looking for to add Former GitHub CEO Nat Friedman and a safe superpower co-founder Daniel Gross. Bansal Meta can help the development of a competitive border AI thinking model with the leading technology of Industrial, Openai’s O3 or Deepseek’s R1. Currently, Meta AI does not openly present the model model.

In recent months, Mark Zuckerberg has been in a company hiring to establish a new AI team of methane, Offers a $ 100 million compensation package to the best researchers connected to the company. It is not known what is offered by the generation to join this agreement.

Nevertheless, it seems that Zuckerberg was the best AI research talent.

The other three past Openai researchers – Lucas Beyer, Alexander Kolesnikov and Xiaohua Zhai – In recent weeks, Meta AI Supertlay team also joined. Wall Street Journal Reported on Wednesday. The former Google Deepmind researcher Jack Rae and former machine learning leader Johan Shearwyk will join them with a former machine teaching leader Bloomberg.

Zuckerberg has been told to further fill the new AI unit Get Startups with heavy strokes AI research laboratoriesSümskhere’s safe Supernellmi, Mira Murati Thought machines and embarrassment of laboratories. However, these negotiations did not exceed the final stage.

The latest podcast, Openai CEO Sam Altman said Meta tried to close the best talent of the beginning, but claimed that he claimed that “None of our best people He decided to put it in. ”

A meta spokesman refused to comment.

AI Thinking models provide the main area for the AI ​​Super Header of Meta to be correct. In the last year, Openai, Google and Deepseek sent a high performer that pushes the limits of what the program can do. To work through problems before answering AI models, AI laboratories have successful in improving the performance of AI laboratories, AI laboratories and real world assignments to use the Sources of AI.

Meta’s AI super sophisticated laboratory, as Google’s Deepmind section, can be a key domestic group that strengthens the products throughout the company. Meta makes an ambitious effort Set up AI agents for business CEO of AI former Salesforce CEO Clara Shih. To build competitive agents, Meta must develop border AI thinking models to strengthen them.

BANSAL AND OTHER KEY IS HOPE TO BE POSITION IN META AI RACE WITH AI AI researchers. It may be difficult to take this into account Openai plans to release the Open AI justification model In the coming weeks – a suggestion that can add more pressure on the open AI offers of the meter.

This story has been updated after publication with more information.