Meta's Applied AI Team Faces Internal Turmoil and Talent Mismatch

Meta's internal "Applied‑AI" unit, created in March 2026 and staffed with roughly 6,500 software engineers, has been described by employees as a "brutal" and "gulag‑like" environment. Workers were forced to join the team or face termination, and a recent internal presentation erupted in chaos, with participants shouting accusations at senior AI managers. Upper‑level executives, including Chief Product Officer Chris Cox, have publicly labeled the AI training program as "brutal" and "insane," while CEO Mark Zuckerberg has admitted mistakes and promised no further mass layoffs for the year.

Analysts say the unrest reflects a deeper structural clash between Meta's platform‑centric business model, which rewards rapid feature shipping and user‑engagement metrics, and the long‑term research cadence required for frontier AI work. Compared with rivals such as Google, which maintain separate research arms, Meta’s talent‑acquisition and reward systems are ill‑suited for deep AI development, risking the loss of top engineers to competitors.