Yann LeCun calls Dario Amodei deluded on AI extinction risk

By Splitfeed EditorialLast updated Oct 1, 2026, 15:30 UTC

On October 1, 2026, Yann LeCun, Meta's chief AI scientist and a Turing Award winner, publicly declared he has 'zero concerns' about the possibility of AI causing human extinction. In the same statement, reported by Fortune, LeCun described Anthropic CEO Dario Amodei as 'deluded' and 'crazy' for holding and publicizing existential risk warnings, and further claimed that Amodei does not understand cybersecurity. The exchange came alongside separate reporting on Amodei's own forecasts about AI replacing human professions, drawing a contrast between two of the most prominent figures in the industry.

The dispute matters because LeCun and Amodei are not peripheral voices. LeCun, often called an AI godfather, represents a school of thought that treats current AI architectures as fundamentally incapable of producing the kind of autonomous, goal-driven behavior that extinction scenarios depend on. Amodei, whose company Anthropic was built partly around safety-first principles, has repeatedly argued that advanced AI poses genuine catastrophic risks. When researchers of this standing attack each other's credibility in public, it signals that the internal disagreement about AI risk is not narrowing as the technology advances, it is sharpening.

None of the three camps had published reactions at the time of writing, so what follows reflects what each would typically argue in a story like this. The Pro-AI camp would likely celebrate LeCun's position as a grounded corrective to what it sees as fear-driven regulation and corporate safety theater. The Anti-AI camp would probably defend Amodei, pointing to Anthropic's research as evidence that caution is warranted and dismissing LeCun's confidence as premature. The Middle Ground camp would tend to argue that both men raise legitimate points, that architectural limits today do not guarantee safety tomorrow, and that personal attacks between researchers obscure rather than resolve the underlying technical questions.

The argument is unlikely to be settled by a single moment, but Anthropic's forthcoming research publications and any formal response from Amodei would be worth watching, as would whether LeCun elaborates his cybersecurity criticism in a technical forum where specific claims could be examined more closely.