Meta’s chief AI scientist Yann LeCun reportedly plans to leave to build his own startup

Meta’s longtime chief AI scientist and Turing Award winner, Yann LeCun, is reportedly preparing to leave the company to start his own venture, according to the Financial Times.

LeCun, who also serves as a professor at New York University and is one of the pioneers of modern deep learning, is reportedly raising capital for a new AI startup focused on advancing his research into world models — AI systems designed to build an internal understanding of their environment and simulate cause-and-effect relationships.

If confirmed, his departure would mark the exit of one of Meta’s most influential AI figures and a potential turning point for the company’s research division.

A Defining Exit for Meta’s AI Ambitions

LeCun joined Meta (then Facebook) in 2013 and helped launch FAIR — the Fundamental AI Research Lab — to focus on long-term, foundational research. While FAIR produced several breakthroughs, its role has recently been overshadowed by CEO Mark Zuckerberg’s push toward short-term generative AI projects.

Sources say Meta’s new Meta Superintelligence Labs (MSL), led by Scale AI CEO Alexandr Wang, has absorbed significant resources and attention after Meta invested $14.3 billion in Wang’s data-labelling company earlier this year. The restructuring, along with the hiring of more than 50 engineers and researchers from rival firms, has reportedly created internal friction and organisational confusion.

At the same time, Meta’s Llama 4 model family struggled to match the capabilities of competitors from OpenAI, Google, and Anthropic, fueling a sense of urgency that some insiders say conflicts with LeCun’s slower, research-driven approach.

LeCun’s Vision for World Models

LeCun has long argued that today’s large language models (LLMs) — including Meta’s own — lack the reasoning and environmental understanding required for actual intelligence. His research into world models aims to fix that, enabling AI systems to predict outcomes and interact with their surroundings more like humans or animals.

He has been openly sceptical of the AI industry’s hype cycle, frequently criticising claims about “superintelligent” systems. In one recent post, he wrote:

“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.”

LeCun’s forthcoming startup would reportedly focus on turning that vision into practice — potentially competing with initiatives at Google DeepMind and World Labs, both of which are developing similar architectures.

An Uncertain Future for Meta’s AI Research

LeCun’s departure would represent a significant loss for Meta’s research culture, which he has helped shape for over a decade. It also underscores the growing divide between scientific AI research and the commercial race to ship generative products.

Meta has not commented publicly on the report.