Yann LeCun steps down as Meta’s chief AI scientist to launch new startup

Yann LeCun is departing Meta after more than a decade, marking one of the biggest shake-ups in the company’s AI leadership this year. 

In a LinkedIn post, LeCun announced that he is launching a new venture dedicated to developing “world model” technologies, i.e., AI systems designed to understand the physical world, maintain memory, and plan complex actions beyond traditional data-trained models.

His new startup will continue the Advanced Machine Intelligence (AMI) programme he pioneered with collaborators over the years. 

Despite the exit, Meta will remain a partner of the project, a sign of the company’s ongoing interest in LeCun’s long-term research goals.

However, the timing is noteworthy as it is arriving just months after major restructuring within Meta’s AI division, which struggled with the lukewarm reception of its Llama-4 model. 

CEO Mark Zuckerberg has since poured billions into rebuilding its credibility, recruiting top researchers and investing heavily in new AI initiatives.

For those uninitiated, Yann LeCun, who joined the company in 2013, leaves behind a legacy rooted in deep learning research and the creation of FAIR, one of the field’s most influential labs. 

As Yann LeCun moves into this new chapter, industry watchers say his startup could reshape the race to build more human-like AI systems.