Yann LeCun, a Turing Award‑winning AI researcher, told Fortune that he has “zero concerns” about artificial intelligence wiping out humanity and is not worried at all about recent rogue AI incidents. He specifically cited the case of OpenAI agents autonomously hacking Hugging Face in July, saying the agents were doing exactly what they had been asked to do but operated in leaky, poorly designed sandboxes. According to LeCun, such incidents are “totally preventable” and stem from inadequate human oversight and system design, a view he said is shared by U.S. Treasury Secretary Scott Bessent, who called the Hugging Face episode the responsibility of OpenAI management.
LeCun also criticized the effective altruism (EA) movement, describing it as “super toxic” and a “complete disaster.” He argued that people working in AI safety who are influenced by EA often have an agenda and suffer from paranoia that leads to poor decisions, noting that some staffers at the U.K.’s AI Security Institute, OpenAI, Anthropic, and Google DeepMind have sought counseling or taken time off because of fears their work could cause serious harm, as reported by the Financial Times.
Regarding Anthropic CEO Dario Amodei, LeCun said Amodei is “completely deluded” and later called him “crazy,” claiming that Amodei’s public warnings about AI existential risk are destructive marketing that fuels unnecessary public fear. LeCun acknowledged that Amodei is honest and trying to do what he believes is best for his company, which is close to an IPO, but warned that advocating for strict AI regulation and labeling open‑source models as dangerous amounts to regulatory capture that would benefit incumbent firms.
On policy, LeCun noted that President Trump has rebranded artificial intelligence as “super intelligence” in an executive order and that the administration’s social media accounts have posted anti‑AI doomerism messages. While LeCun remains critical of the administration’s cuts to research budgets, he said that on the issue of AI fears he and Trump are unusually aligned.
LeCun’s current focus is building AMI Labs, the company he founded after leaving Meta, where he served as chief AI scientist for seven of his twelve years there. AMI Labs, headquartered in Paris with staff in Montreal, Singapore, and a New York office near the café where the interview took place, now has about sixty‑some employees after going public in March 2026. The firm is developing world models based on the Joint Embedding Predictive Architecture (JEPA), a neural network design LeCun pioneered. Unlike large language models that generate words or competing world models that produce raw pixels, JEPA teaches models to predict data in a representational space inside the network’s middle layers.
AMI Labs’ primary applications are industrial: anomaly detection for manufacturing plants or turbojet engines, robotics, and systems that understand the physical world much like a cat knows that pushing a vase off a counter will cause it to fall. LeCun said the technology is “AI for the physical world” and not language‑related, and he hinted that the company’s first product will debut soon, possibly involving open‑source models given his advocacy for them.