Yann LeCun founded Meta AI in 2013 as a Facebook artificial intelligence research lab, or FAIR, and with Meta CEO Mark Zuckerberg. For many years, Meta AI was the point of reference for millions of researchers worldwide. FAIR built Meta's reputation as an open academic-style AI research lab producing work like PyTorch, computer vision breakthroughs, and self-supervised learning research that were also praised by industry leaders. 'Let's face it, the single most important thing that has happened last year, if you were to ask me the one single most important event last year and how it has activated AI researchers here in this region, it's actually Llama 2.' Over time, Meta AI became the umbrella name for the company's AI efforts, which included research products like the Llama models, the Meta AI assistant in WhatsApp, Instagram, Messenger, the Ray-Ban Meta Glasses, and all the related infrastructure. FAIR remained the core research arm inside the Meta AI umbrella, or in other words, the pure research division as opposed to product-focused teams. In May 2017, Joelle Pineau joined Meta AI and became vice president of AI research. Under her leadership, FAIR helped drive Meta's Llama development and expanded Meta AI's efforts to make cutting-edge AI models openly available to researchers and developers. During the second semester of 2025, things started to change. Joelle Pineau left Meta. In June, Meta invested $4.3 billion for a 49% stake in Scale AI and brought in Scale AI's CEO Alexander Wang to Meta as its first ever chief AI officer. Then in 2025, also Yann LeCun left Meta to found his own venture, Advanced Machine Learning Labs, or AMI, pursuing an alternative architecture to large language models. LeCun went silent for a while, but he came back recently at VivaTech 2026 and explained exactly why he left Meta.
The project I was working on has been my obsession for a long time, at least 15 years, but more intensely over the last five. Internally at Meta, we called it AMI, Advanced Machine Intelligence. That became the name of the company, but it was the name of the project internally at FAIR. This project had the support of the leadership of FAIR. Big secret is I was not the leader of FAIR. I was, you know, a researcher essentially. Yeah. It was led by Joelle Pineau until she left last year. So it was an important project for FAIR. It had the support of Mark Zuckerberg, Andrew Bosworth the CTO, Mike Schroepfer the former CTO. The leadership was supporting it. But what happened last year is that the company chose to refocus its effort, including some of its research efforts, towards LLM and catching up with the rest of the industry. Still willing to support my project, but the general ambience was not as favorable, even though I had the support of the leadership of the company, and we realized also that our techniques were starting to work really well. Most of the applications were in industry, and those are domains that Meta is not very interested in. Meta is really about connecting people with each other. That's the only thing they care about, at least for now. And so probably it was, you know, it made sense to just leave the company, try to accelerate the development, and build the things we can build with it. And I realized I could also raise enough money to keep this going. There was a series of conversations, you know, starting late 2024. When I, there was sort of a bit of a conundrum because the GenAI organization which was working on Llama was really focused on very short-term objectives, and to some extent there was a bit of a distance and impetus mismatch with FAIR which was really about fundamental research. So I told Mark that we should probably have an organization in between that sort of bridges the gap between the two things, and that would give the opportunity to FAIR to really work on ambitious longer-term projects. One of those two recommendations actually occurred. So Mark in 2025 created this intermediate organization which is called TBD Lab. But then what happened to FAIR is that FAIR was actually pushed into contributing to TBD Lab, actually refocusing on shorter-term objectives, and that went against what I thought. There was also ideas about being less open and putting restrictions on publications which I disliked. So it was a combination of things really, but there's no, you know, I'm not angry. I mean, I think there was long discussions after the release of Llama 1 which was not really open. Llama 2 was being worked on, and the question was: should Llama 2 be open source? Internal discussions, you know, with Mark Zuckerberg and 40 people down from the leadership every week debating that topic. There's really real questions there. At one point, I was arguing vehemently for open sourcing Llama 2, and so did Andrew Bosworth. I mean, it's not like I was the only one, but certainly I was vocal on this. And at some point Mark said, 'Yeah, we're going to open source Llama. Tell me how.' And that was a watershed moment because when Llama 2 was open sourced, a lot of people realized, 'Oh, wow. We can build a whole industry around this, right?' And a lot of startups really were made possible by the open sourcing of Llama. And it's a bit sad. I think that now the ethos has changed but, okay, it's competitive. A lot of people, particularly people who are close to the US government, which for all of its characteristics is actually somewhat in favor of open source AI, and they see that American industry is not doing open source anymore, and that the open source scene is basically entirely occupied, at least at the frontier, by Chinese companies. And what you see in the utilization of AI platforms, there's a huge increase in the use of Chinese models which are pretty good and they're cheap, and you know, cheap to run.
They're free if they're open. So so that creates kind of a conflict. And in the middle of this you have companies like Anthropic and a few others that are basically lobbying to essentially make open source illegal, right? Because they think AI is intrinsically dangerous. I've had this idea for a number of years. I tried to push it inside Meta with very limited success. I've been trying to advertise it in various forums and governments, and I've been talking to a bunch of governments around the world. They all want AI sovereignty, right? And I think they're right. It's very important because pretty soon all of our information diet will be mediated by AI assistants. I'm actually wearing one on my nose right now. The Meta, you know, these are the smart glasses. I can take a picture of you guys, right? Okay. Smile.
