Meta is stealing talent with its biggest rivals as it is ready to become a serious contender in the AI race. And now, after keeping a team together, CEO Mark Zuckerberg has announced that Shengzia Zhao, one of the co-writers of a former OpenEI researcher and the original chatgipt paper, has been named Chief Scientist of the new mining superintendent AI group of Mata.

Jhao, who quietly joined the Meta in June, was playing an important role in Openi’s early successes, from the first Reasoning Model, the creation of O1, to help the shape chat. That O1 model, coincidentally, is very one who launched the “chain-off-thot” craze, which was later picked up by the choice of Google and Deepsek.

Meta CEO Mark Zuckerberg of Meta confirmed the promotion of Zhao in a post on threads, called him “from our leading day”. He explained, “Now that our recruitment is getting good, and our team is coming together, we have decided to formalize their leadership role.”

Zhao will now work directly under Alexandra Wang, who was the CEO of AI on the pre -scale AI, who joined Meta as his Chief AI officer in June. Wang has been assigned anything less than the steering meta towards the Artificial General Intelligence (AGI), which the systems are able to argue and think at human or even supernatural levels.

The Superintending Lab, which was unveiled by Meta in June 2025, is the focal point of this new ambition. It operates separately from the long -established AI Research Unit of Meta, Meta, led by AI veteran Yan Lakeun. Lakun now reports to Wang, which gives a clear inspection of Meta’s two-dimensional AI Research Empire.

Meta’s AI hiring biping has been in the headlines in all summer. In a few months, the company has wooed more than a dozen leading researchers of Openai, Apple, Google and Ethropic. This includes high-profile names like Apple AI scientists Tom Gunter and Mark Lee.

Meta’s high-stake AI talent did not stop with Hunt Shengzia Zhao. As information was first given in June, Zhao three other influential Openi joined the researchers: Jiahui U, Shuchao B, and Hongyu Rain. Tech veteran has also brought a trapit Bansal, who had previously collaborated with Zhao on the AI Reasoning Model, and recruited a trio of engineers from the Jurich office of Openai, each, each, had specializing in multimodality.

To do this, CEO Mark Zuckerberg has personally rebelled the recruitment, allegedly reach the candidates via email and even host possible rent in his private lake Taho Retreat. Your average job is not an interview venue. According to reports, some offers have been generously generous with eye water, with compensation packages reaching eight and nine figures. And Meta is not playing all around, many of these are “explosions offers”, designed for top researchers to sign within a few days.

Currently, the Meta’s most advanced open-source model, Lama 4, is still behind rivals such as GPT4 of Openai and Gemini of Google. But the company is placing a big bet on a new model, named “Bhamath”, which is expected to be a debut at the end of this year. Zuckerberg saw what is coming next. “Together we are building an aristocratic, talented-gray team that focuses resources and long-term focus to pursue the Superintending Research front,” he said.

With Zhao, now with the driver’s seat and the growing team of AI Heavyweight, Meta wants to clearly ensure that the next major success in AI has approved its logo.