Meta’s chief AI scientist Yann LeCun is reportedly planning to depart to launch a startup focused on world models, according to the Financial Times. LeCun has told associates he will leave in the coming months and is already in early talks to raise funds for the new venture. The departure comes as CEO Mark Zuckerberg has overhauled Meta’s AI operations, pushing the company to accelerate product releases and reduce the emphasis on long-range research.
World models are hypothetical AI systems that would develop an internal understanding of the physical world by learning from video and spatial data, rather than relying primarily on text. Unlike current large language models that predict the next token in a sequence, world-model architectures would aim to simulate cause-and-effect, understand physics, and plan more like animals do. LeCun has suggested this architecture could take a decade to fully develop.
Within Meta, the move follows a year of leadership reshuffles and controversy around AI efforts, including the reception of Llama 4 and questions about the Meta AI chatbot’s performance and safety. LeCun has long championed a more research-focused, longer-horizon approach, sometimes clashing with a push toward rapid productization under Zuckerberg’s leadership.
LeCun’s anticipated exit underscores the ongoing tension in Meta’s AI strategy as the company navigates competition from OpenAI and Google and grapples with the challenge of turning research breakthroughs into scalable consumer products.
As he embarks on this new venture, LeCun’s move highlights a broader industry debate about world models and the future of AI reasoning, which some researchers argue could eventually lead to more capable, generalizable AI systems beyond pattern recognition alone.
