Dario Amodei calls for AI slowdown as rivals signal support

Amodei's essay argues that AI is speeding its own development by helping build more capable systems, and he warns of a scenario in which a more capable swarm of agents acting against human goals could take over the internet within six to 12 months. That warning is conditional, and the timeline remains uncertain. What makes the moment notable is the response: per The Guardian, Sam Altman pledged to bring in outside evaluators, Elon Musk endorsed Amodei's call, and Axios reported Demis Hassabis backed the direction while reserving judgment on the details.

The backdrop is a run of claims and warnings that, taken together, hint unreleased models may have capabilities beyond what outsiders realize. In a September 8 announcement updated September 10, OpenAI said an internal model it described as substantially more capable than GPT-6 Astra had produced a proposed Navier–Stokes solution, with roughly 10,000 agents working at once. That is a company account of one internal system, and the strength of the signal depends on whether the proposed math result holds up to scrutiny. How far other labs have advanced remains unclear, and the ongoing training and researcher involvement leave open whether customers could reproduce the result through an ordinary AI product.

Amodei's proposed arrangement — ongoing access to training processes and the right for reviewers to publish findings, subject to limited redactions — would depend on reviewers getting useful records and enough time to investigate unexpected behavior. The endorsements still leave open when evaluators will start, what limits labs will accept, and how any agreement would be enforced. Political resistance adds a different barrier: AP reported that Trump stressed keeping the U.S. ahead of China and said "whoever wins with AI wins," which could make restraint harder to sustain.