Meta has appointed Robert Fergus, a former director at Google DeepMind, to lead its Fundamental AI Research (FAIR) division, according to a report from Bloomberg.
The decision comes as Meta continues restructuring its artificial intelligence operations following a wave of leadership changes and internal talent shifts.
Fergus, who previously served as a research scientist at Meta before joining DeepMind, brings years of experience leading AI research. During his five-year tenure at DeepMind, he was involved in several advanced research initiatives. His return to Meta marks a strategic move aimed at revitalizing FAIR, a division that has played a foundational role in Meta’s AI advancements.
Founded in 2013, FAIR was responsible for early breakthroughs including the development of Meta’s Llama 1 and Llama 2 language models. However, the lab has faced internal setbacks in recent years, with many researchers reportedly leaving for startups, competitors, or other Meta teams. Some of that talent migrated to Meta’s Generative AI (GenAI) unit, which recently spearheaded the launch of Llama 4 and is increasingly viewed as the company’s main engine for product-focused AI development.
Fergus’s appointment follows the recent departure of Joelle Pineau, Meta’s former vice president of AI research, who announced her exit in April. Pineau played a key leadership role in shaping Meta’s research agenda and expanding its global AI efforts.
By bringing Fergus back to lead FAIR, Meta signals a renewed commitment to its foundational research initiatives alongside its product-driven GenAI work. The move reflects Meta’s broader strategy to balance scientific research with applied innovation in the competitive AI landscape.
Fergus is expected to refocus the FAIR lab’s efforts and help align its goals with Meta’s long-term ambitions in artificial intelligence. While no further organizational changes have been announced, the company’s decision to tap a high-profile AI veteran suggests a push to strengthen its leadership bench and retain relevance in the evolving field of machine learning and generative models.
