Turnover Hits Meta’s New AI Division as Early Hires Exit

Meta’s new AI division, Superintelligence Labs, is undergoing significant personnel and organizational changes less than two months after its creation.

Formed in June 2025 as part of a broader strategy to accelerate AI development, the division has already seen the departure of at least eight employees, including experienced engineers and researchers, as well as recent hires.

Some of those who left had played key roles in building Meta’s AI infrastructure.

Others chose to return to their former employers after a brief time at the company.

The departures come as Meta continues to restructure its AI teams and invest heavily in research and development. The company has emphasized that attrition is common at this scale and pointed to its ongoing hiring efforts. Still, the combination of turnover and early internal changes may suggest the new division is still in the process of finding its footing.

Why It Matters: Meta is competing with companies like OpenAI and Google to lead in AI research and development. The early changes in staffing and structure at Superintelligence Labs highlight the challenges of scaling AI efforts quickly in a competitive and high-profile environment. These developments could influence how the company approaches future AI initiatives and talent strategy.

• Experienced Employees Have Left Since June: Several experienced engineers who had spent years at Meta left within weeks of the Superintelligence Labs being announced. Bert Maher, a twelve-year veteran involved in building PyTorch and Triton, joined Anthropic. Tony Liu, who managed teams working on GPU infrastructure for model training, also stepped away after more than eight years at the company. Both had been part of Meta’s core AI development efforts. Their departures suggest a shift in Meta’s technical leadership during a period of structural change.

• Some Recruits Returned to Previous Employers: Meta hired several researchers from OpenAI during its recent recruiting campaign, but not all stayed. Avi Verma reportedly withdrew before his official start date. Ethan Knight, another researcher hired from OpenAI, left after less than a month. Both have since returned to their previous roles. Chaya Nayak, who spent nearly a decade at Meta, also announced a move to OpenAI.

• Reorganization Followed Initial Launch: Meta restructured the Superintelligence Labs into four teams within weeks of its announcement. This adjustment came after a series of earlier reorganizations within the company’s broader AI division. The latest changes reflect an effort to clarify roles and priorities within the new lab, but they have also contributed to a sense of ongoing transition.

• Workplace Dynamics May Have Influenced Exits: Meta’s push to attract AI talent from competitors included large compensation offers, sometimes reaching into nine-figure packages. While this approach brought in new hires, it also appears to have created tension with existing employees who had been with the company for years. Several former staff described internal dynamics as uneven, with unclear communication around shifting goals and team structures.

• Meta’s AI Work Continues Amid Public Scrutiny: In parallel with internal changes, Meta has faced public questions about its AI tools. Government agencies have opened investigations into chatbot performance and moderation. Meta has not publicly linked these issues to employee departures.

Go Deeper -> Meta Superintelligence Labs is already losing key staff less than two months after launch – Business Insider

Meta’s Already Bleeding AI Talent Two Months Into Hiring Spree – Gizmodo