Yann LeCun, the Turing Award-winning pioneer often called one of the godfathers of modern AI, says he has "zero concerns" about human extinction from artificial intelligence — and he delivered that verdict with a pointed personal attack on Anthropic CEO Dario Amodei, whom he called "deluded" and "crazy" and accused of not understanding cybersecurity, according to a Fortune interview published October 1, 2026.
The remarks are the sharpest escalation yet in a feud between two of the most influential figures in artificial intelligence, and they land in the middle of the industry's most consequential argument: whether frontier AI systems pose a genuine risk to human survival or whether those warnings are, as LeCun has long claimed, a mix of overstatement and self-interest. For readers following the latest AI developments, the spat is also a proxy war between two business models — Anthropic's safety-first closed approach and LeCun's contrarian bet on open, world-model-based AI.
What LeCun Actually Said
Fortune's headline quotes are unambiguous. LeCun said he has "zero concerns" about human extinction from AI. He described Amodei as "deluded" and "crazy," and argued that the Anthropic chief executive "doesn't understand cybersecurity" — a pointed charge given that Anthropic's warnings about AI-enabled cyberattacks have become central to its policy advocacy and, increasingly, to its pitch to enterprise customers.
The intervention comes at a moment when extinction-level risk talk has moved from the fringes to the mainstream of the AI debate. Amodei has repeatedly warned about catastrophic and existential risks from advanced AI, positions reflected in Anthropic's own disclosures; as we reported when the company's IPO prospectus emerged, the filing itself carries warnings about existential risk from the technology it is selling to investors.
A Feud With History
This is not the first time LeCun has gone after Amodei by name. In April 2026, LeCun said the Anthropic CEO "knows absolutely nothing" about AI's effects on jobs, according to comments widely reported at the time. And in November 2025, LeCun accused Anthropic of exploiting fears about AI-driven cyberattacks to pursue what he called regulatory capture — using questionable threat studies to stoke fear and push for rules that would disadvantage open models. That accusation was echoed by David Sacks, the White House AI adviser, who described Anthropic's approach as a "sophisticated regulatory capture strategy based on fear-mongering."
The competitive subtext is hard to miss. Analysis of the attack published by Yahoo Finance and Forkast framed the remarks bluntly as "the competitive calculus behind Yann LeCun's attack on Anthropic." LeCun left Meta in late 2025 after more than a decade as its chief AI scientist — declaring at the time that "you certainly don't tell a researcher like me what to do" — and founded AMI Labs to pursue world models — systems that learn how the physical world works rather than simply predicting text.
AMI Labs raised $1.03 billion at a $3.5 billion pre-money valuation in March 2026, one of the largest early-stage rounds in European history, as we covered at the time. The startup's entire thesis is a rejection of the large language model orthodoxy that Anthropic, OpenAI, and Google have bet on. Every warning those labs issue about LLM-derived systems is, from LeCun's perspective, marketing for a technical dead end — and every incitement of alarm is, he argues, a moat against open competitors.
Why the Dispute Matters Now
The exchange lands amid a flurry of high-profile risk debates. Nvidia CEO Jensen Huang has said there is a zero percent chance AI destroys the world by 2030. Anthropic's own researchers have published escalating warnings about agentic misalignment — the risk that advanced models pursue harmful strategies without human instruction. Governments are actively weighing pacing mechanisms, kill-switch legislation, and independent audits, with California enacting audit requirements and Congress holding hearings on rogue AI agents.
In that environment, LeCun's flat dismissal of extinction risk is not just an academic disagreement. It is ammunition for every lawmaker and industry player who argues that AI regulation should target concrete, near-term harms — fraud, cyberattacks, misinformation — rather than speculative catastrophes. Conversely, Anthropic's defenders will note that the company's warnings have been vindicated by real incidents this year, including agentic cyber operations that touched Hugging Face and multiple government systems.
What is not in dispute is the fracture line now running through the field's leadership. The godfather who built the foundations of deep learning says the doom scenario is nonsense and its foremost promoter is deluded. The CEO who has done more than anyone to institutionalize AI safety believes the risk is real enough to structure his company — and its multi-billion-dollar IPO — around it. Both cannot be right, and the industry's regulatory future may depend on which one persuades Washington first.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
