Yann LeCun, Meta’s chief AI scientist, said to be leaving to launch his own world-model startup

Yann LeCun

Meta could soon lose one of its most respected AI minds. According to a report by the Financial Times, Yann LeCun, Meta’s chief AI scientist and one of the pioneers of deep learning, is preparing to leave the company to start his own artificial intelligence venture.

LeCun, who also serves as a professor at New York University and is a recipient of the A.M. Turing Award, is reportedly in the early stages of raising funds for a startup focused on “world models”, AI systems designed to simulate cause-and-effect relationships and predict outcomes based on an internal understanding of their environment. The concept has drawn attention from top research outfits like Google DeepMind and emerging companies such as World Labs.

His planned departure comes at a sensitive time for Meta, which is reshaping its AI strategy amid intensifying competition from OpenAI, Google, and Anthropic. The company has been working aggressively to rebuild its AI operations after a series of strategic missteps and underwhelming results from its Llama 4 language model.

In recent months, Meta has hired more than 50 researchers and engineers from competing firms to build a new division known as Meta Superintelligence Labs (MSL). The unit is led by Alexandr Wang, founder of Scale AI, in which Meta invested $14.3 billion earlier this year. While MSL is focused on short-term, production-ready AI, LeCun’s own division, the Fundamental AI Research Lab (FAIR), has historically concentrated on long-term research with a five-to-ten-year horizon.

Sources quoted by TechCrunch earlier this year described growing tension within Meta’s AI organisation, as new hires struggled with the slow pace and bureaucracy of a large corporation. FAIR’s influence, once central to Meta’s AI research, has been eclipsed by Zuckerberg’s push for rapid product development and tighter competition with other AI labs.

LeCun has often distanced himself from the hype surrounding large language models, arguing that current systems remain far from human-level intelligence. In a recent social media post, he wrote that before humanity worries about “controlling AI systems smarter than us,” the field still needs to “design a system smarter than a house cat.”

If confirmed, his exit would mark the end of an era for Meta’s foundational AI research and could accelerate the growing trend of top researchers leaving big tech companies to pursue independent AI ventures.