Yann LeCun announces departure from Meta

Yann LeCun, Meta’s chief AI scientist and a co-recipient of the 2018 Turing Award, confirmed he was leaving the company to found a startup working on “world models” — systems intended to build an internal understanding of the physical world rather than predict text. Reporting identified the venture as Advanced Machine Intelligence Labs, co-founded with Alexandre LeBrun.

LeCun had led Meta’s Fundamental AI Research group, FAIR, since 2013, and had argued publicly and repeatedly that large language models were an architectural dead end incapable of the kind of grounded reasoning and planning that world models could provide. That position had put him increasingly at odds with Meta’s direction: the company had folded FAIR into the newly created Meta Superintelligence Labs earlier in 2025, prioritising large-scale LLM development after the underwhelming reception of the Llama 4 family, and reporting on the departure cited friction between LeCun’s long-horizon research agenda and the unit’s more commercially driven mandate under chief executive Mark Zuckerberg.

The departure was reported in advance of LeCun’s own confirmation — the Financial Times and TechCrunch had described the move as likely more than a week earlier — and framed it as part of a broader churn of senior researchers leaving large labs to pursue world-model and alternative-architecture research with outside capital, rather than inside the LLM-centric roadmaps that had come to dominate frontier labs’ spending. LeCun said he intended to begin fundraising for the new venture.