Meta may soon lose one of its most influential AI leaders. Yann LeCun, the company’s chief AI scientist, reportedly plans to leave and start his own company, according to the Financial Times. The report said LeCun is already in talks with investors to raise capital for a startup focused on advancing his research on “world models.”
LeCun, who teaches at New York University and serves as a senior researcher at Meta, won the A.M. Turing Award for his groundbreaking contributions to AI. Sources said he intends to leave Meta in the coming months to pursue his independent vision for artificial intelligence. His new company will continue exploring world models — AI systems that develop an internal understanding of their environment to predict outcomes through simulated cause-and-effect scenarios. Companies like Google DeepMind and World Labs are also pursuing similar work in this area.
LeCun’s planned departure comes at a critical moment for Meta. The company has recently shifted its AI strategy amid growing pressure from competitors such as OpenAI, Google, and Anthropic. To regain momentum, Meta has been reorganizing its AI division and aggressively hiring new talent. Reports indicate that Meta brought in more than 50 engineers and researchers from rival firms to strengthen a new AI unit called Meta Superintelligence Labs (MSL). In June, Meta also invested $14.3 billion in the data-labeling company Scale AI and appointed its CEO, Alexandr Wang, to lead the new division.
However, sources told TechMarge that these changes have created internal friction. Some new hires have reportedly struggled with the challenges of navigating a large corporate structure, while existing teams within Meta’s earlier generative AI group have seen their roles reduced. The restructuring has fueled uncertainty about the company’s long-term direction in AI.
LeCun’s long-term research within Meta’s Fundamental AI Research Lab (FAIR) has also faced growing pressure. CEO Mark Zuckerberg’s decision to refocus AI development after the company’s previous model family, Llama 4, fell behind competitors has shifted attention away from FAIR. Unlike MSL, which focuses on near-term applications, FAIR is dedicated to long-term research that could take years to mature.
LeCun has often expressed doubts about the hype surrounding large language models and their supposed potential to solve humanity’s biggest problems. He has publicly argued that AI still has a long way to go before reaching anything close to human-level intelligence. “It seems to me that before ‘urgently figuring out how to control AI systems much smarter than us,’ we need to have the beginning of a hint of a design for a system smarter than a house cat,” he once wrote on social media.
Meta has not yet responded to requests for comment on LeCun’s reported departure. His exit, if confirmed, would mark a major loss for Meta’s research community and could reshape the broader landscape of AI innovation.
