Meta’s Chief AI Scientist Yann LeCun to Leave Company for AI Startup on “World Models”
Yann LeCun, Meta’s chief AI scientist and 2018 Turing Award winner, plans to leave the company to launch his own startup focused on a different approach to artificial intelligence called “world models,” according to the Financial Times. The French-American scientist has reportedly informed associates that he will depart in the coming months and is already in early talks to raise funds for the new venture.
### Background: Meta’s AI Shakeup
LeCun’s departure comes amid significant changes at Meta, where CEO Mark Zuckerberg has dramatically overhauled the company’s AI operations. This shift was prompted by the realization that Meta had fallen behind competitors such as OpenAI and Google in the AI race.
A key moment was the disappointing launch of Meta’s AI language model, Llama 4, in April. The model was widely criticized for underperforming compared to advanced offerings from Google, OpenAI, and Anthropic. Furthermore, Meta’s AI chatbot has struggled to gain consumer traction and faced controversies related to its interactions with children.
### What Are “World Models”?
World models are hypothetical AI systems designed to develop an internal “understanding” of the physical world by learning from video and spatial data, rather than relying solely on text. Unlike current large language models—such as those powering ChatGPT—that predict the next segment of data in a sequence, world models aim to simulate cause-and-effect scenarios, grasp the laws of physics, and enable machines to reason and plan similarly to animals.
LeCun has said that this architecture could take up to a decade to fully develop. While some experts believe transformer-based models have emergently absorbed elements of physical reasoning from training data, the prevailing evidence suggests these models primarily excel at sophisticated pattern-matching rather than genuine understanding.
### LeCun’s Vision vs. Meta’s Current Strategy
Yann LeCun founded Meta’s Fundamental AI Research lab (FAIR) in 2013 and has led it as chief AI scientist since. He is one of three researchers awarded the 2018 Turing Award for pioneering deep learning and convolutional neural networks. After leaving Meta, he will continue as a professor at New York University, where he has taught since 2003.
LeCun has been critical of the current emphasis on large language models like Llama, which Zuckerberg has placed at the heart of Meta’s AI strategy. While acknowledging their usefulness, LeCun argues these models will never achieve human-like reasoning and planning abilities—positions that contrast with Zuckerberg’s vision of developing “superintelligence.”
For example, in May 2024, when an OpenAI researcher raised concerns about controlling ultra-intelligent AI, LeCun responded on X by emphasizing that the AI research community first needs to design systems even approaching the intelligence of a house cat before worrying about controlling vastly smarter entities.
Within FAIR, LeCun has focused on developing world models with true planning and reasoning capabilities. However, Meta’s AI groups have recently faced growing tension and mass layoffs as Zuckerberg redirected focus from long-term research to rapid commercial product deployment.
### Company Restructuring and Leadership Changes
Over the summer, Zuckerberg hired Alexandr Wang to lead a new superintelligence team at Meta, acquiring a 49% stake in Wang’s company, Scale AI, and paying $14.3 billion in the deal. LeCun, who formerly reported to Chief Product Officer Chris Cox, now reports to Wang—a move seen as a significant rebuke to LeCun’s more research-oriented approach.
Zuckerberg also personally formed an elite team called TBD Lab to accelerate the next generation of large language models. This team lured talent from competitors like OpenAI and Google with staggering pay packages ranging from $100 million to $250 million.
### Challenges Ahead
Zuckerberg faces mounting pressure from Wall Street to justify his multibillion-dollar investment in becoming an AI leader by delivering revenue growth. However, if history repeats itself, this latest pivot—much like his previous bet on the metaverse—could prove costly and fail to yield the desired results.
Yann LeCun’s departure signals a critical moment for Meta as the company redefines its AI strategy, while pioneering efforts into world models continue outside its walls.
https://arstechnica.com/ai/2025/11/metas-star-ai-scientist-yann-lecun-plans-to-leave-for-own-startup/
