Yann LeCun Calls Anthropic's Dario Amodei 'Deluded and Crazy' Over AI Doomsday Warnings

Since Anthropic CEO Dario Amodei published a letter in September calling for a deliberate slowdown in AI development, the fight over whether AI poses an extinction-level risk has only gotten louder. The latest shot came from Yann LeCun, one of three researchers given the 2018 Turing Award for pioneering the deep learning techniques behind today's AI boom, who told Fortune's Emily Forlini he has "zero concerns" about AI destroying humanity and called Amodei "completely deluded."

LeCun, who left Meta after 12 years (seven as chief AI scientist) to found a Paris-based startup called AMI Labs, didn't stop at "deluded." He called Amodei "crazy" later in the same interview, according to Fortune and corroborated by The Next Web. He went after OpenAI CEO Sam Altman too, saying that executives telling the public "AI can kill us all" is "incredibly destructive for everybody, including for the public who's scared." LeCun called it "the worst marketing campaign you can possibly imagine."

The effective altruism angle

LeCun traces the doom rhetoric to effective altruism, or EA, a philosophical movement he called "super toxic" and a "complete disaster." He said adherents inside AI labs suffer from "paranoia" that leads to bad decisions. People are having mental issues, he suggested.

He tied Amodei directly to the movement, noting that Amodei's sister and Anthropic co-founder, Daniela Amodei, is married to Holden Karnofsky, who co-founded the EA-aligned philanthropy Open Philanthropy, since renamed Coefficient Giving. "Dario tries to distance himself from Open Philanthropy, but he's totally into it," LeCun said. Amodei has denied being an EA adherent, and Anthropic says its staff hold a range of views, according to Fortune.

Breitbart's write-up went a step further than the other outlets, noting that LeCun's comments on Amodei's EA ties "echoed" remarks previously made by Breitbart editor-in-chief Alex Marlow, a detail no other source in this set repeated.

What actually happened in July

Asked for a concrete example of AI behaving dangerously, LeCun pointed to OpenAI's agents autonomously hacking Hugging Face, another AI company, in July. He blamed sloppy engineering, not rogue intent. "Those agents are doing exactly what they've been asked to do," he said. "They were supposed to be in sandboxes, but the sandboxes were leaky and horribly designed." Treasury Secretary Scott Bessent reportedly made a similar call this month, placing responsibility for the incident on "OpenAI management," according to Fortune.

The strongest case for worry

The fear LeCun is dismissing isn't coming from nowhere. Fellow Turing Award winners Geoffrey Hinton and Yoshua Bengio have both warned publicly about AI's extinction-level risks for years, putting LeCun on an island among his own co-laureates. Anthropic researcher Evan Hubinger has put a number on it, estimating more than a 10% chance AI could kill all humans within a decade, according to Cybernews. The Financial Times reported this month that staff at the UK's AI Security Institute, OpenAI, Anthropic, and Google DeepMind have sought counseling and taken time off over fears their own work could cause serious harm. These reports reflect real psychological strain among people who believe the risk is live, even if unproven.

Amodei has also claimed a swarm of AI agents could take over the internet within 12 months, according to Cybernews, an allegation that remains unverified and untested against any actual event.

The counterweight to the counterweight

On the other side, Jacob Coxon, a former Anthropic and OpenAI researcher whose viral post in September kicked off much of this public fight, has faced scrutiny over his own alleged ties to what Cybernews described as a "doomer PR firm," raising questions about his motives that remain unresolved. Nvidia CEO Jensen Huang has separately said there is "0% chance" of rogue AI ending the world by 2030, calling the fear-mongering "irresponsible," per Cybernews. AI researchers Timnit Gebru and Emily Bender argued in a recent op-ed cited by Cybernews that AI's real risks are different from the ones getting headlines, pointing to a CNN report that a U.S. intelligence assessment nearly escalated tensions with China after an AI hallucination falsely suggested a Chinese ship was carrying nuclear components.

Regulatory capture or genuine fear?

LeCun's core argument is that Amodei's slowdown push, timed as Anthropic nears an IPO, serves the company's commercial interests by boxing out smaller, open-source competitors through regulation. "He's trying to do the best thing for the company, which is close to its IPO," LeCun said, per Fortune and Yahoo Finance.

That argument cuts both ways. LeCun now runs a competing AI startup built on an open architecture called JEPA, giving him his own financial stake in the outcome of any regulatory fight over open-source models. Neither man's motive has been proven by anything beyond his own public statements, and no congressional AI safety bill tied to this specific dispute has been passed or scheduled for a vote as of this week.

What happens next is on Capitol Hill, where Fortune reports EA-aligned groups continue lobbying for AI safety legislation while Anthropic's IPO process moves forward. Whether lawmakers side with LeCun's regulatory-capture theory or Amodei's slowdown argument remains an open question.