Person
Yann LeCun
Yann LeCun is a French-American computer scientist who shared the 2018 Turing Award with Geoffrey Hinton and Yoshua Bengio for the deep-learning work that underpins modern AI. He built and led Meta's Fundamental AI Research group, FAIR, from 2013, and became the field's most prominent sceptic of two ideas widely held elsewhere: that scaling up large language models is a route to general intelligence, and that today's systems pose a near-term risk of human extinction. He argues instead for "world models" — systems that learn how the physical world behaves rather than predicting text — an approach he set out in a 2022 paper on the architecture he calls JEPA. That disagreement with Meta's strategy led him to leave the company in late 2025 and, in March 2026, to launch a startup, AMI Labs, built around the world-model idea.
Appears alongside
Featured in threads
Tracks
- Models & capabilities 6
- Open weights & ecosystem 4
- Ideas & essays 3
- Labs & people 2
- Benchmarks & progress 1
- Culture & impact 1
Yann LeCun launches AMI Labs to build JEPA-based world models
The $1.03 billion seed round, at a $3.5 billion pre-money valuation, was co-led by Cathay Innovation, Greycroft, Hiro Capital and HV Capital, with Jeff Bezos, Nvidia and Temasek among the backers.
Labs & people · Models & capabilities
Yann LeCun announces departure from Meta
Meta chief AI scientist Yann LeCun, who built FAIR in 2013, said he is leaving to found a world-models startup, reportedly Advanced Machine Intelligence Labs.
Labs & people
Llama 4 lands badly
Meta's mixture-of-experts release was undercut by accusations that a version tuned for LMArena differed from the public weights.
Open weights & ecosystem · Benchmarks & progress · Models & capabilities
Meta releases V-JEPA
Trained without labels, the model learned by predicting masked video regions in representation space, distinguishing fine-grained actions over roughly ten-second clips.
Open weights & ecosystem · Models & capabilities · Ideas & essays
Meta releases I-JEPA
Trained on unlabelled images by predicting abstract representations rather than pixels, the 632-million-parameter model reportedly matched state-of-the-art low-shot accuracy using far less compute than rival methods.
Open weights & ecosystem · Models & capabilities · Ideas & essays
Meta releases LLaMA to researchers, and it leaks within a week
Meta shared a competitive foundation model with approved researchers; the weights appeared on BitTorrent days later and an open ecosystem formed around them.
Open weights & ecosystem · Models & capabilities
Meta pulls Galactica after three days
Its errors were formatted exactly like real citations and papers, so only an expert reader could tell fabrication from fact — a distinct failure mode from earlier chatbots' obvious mistakes.
Models & capabilities · Culture & impact
Yann LeCun publishes 'A Path Towards Autonomous Machine Intelligence'
The Meta chief scientist proposed non-generative 'joint embedding predictive architectures' as a route to machine intelligence, arguing autoregressive LLMs were a dead end for reasoning and planning.
Ideas & essays
Also mentioned in 2 entries
Referenced in passing — Yann LeCun isn't the main subject of these.