Amodei's AI slowdown call wins backing from Altman and Musk

Amodei's essay is not a call to halt work. He argues the leading developers should leave more space between major capability advances so new models can be evaluated properly before release. To make that concrete, he laid out a three-step approach: firms welcoming independent third-party evaluators with extensive access — a step Anthropic says it has already committed to — then cooperation across the industry, then some form of international agreement so that other countries, notably China, also agree to pace development.

The responses came fast. Musk replied to Amodei's post on X with "Dario is right," and Altman wrote that he agrees "we need to pace the frontier," adding that OpenAI would welcome independent evaluators with employee-like access and would "do the same." Altman also said the topic had been a primary subject of internal discussions at OpenAI in recent weeks, and OpenAI chief scientist Jakub Pachocki had recently proposed a voluntary pause or slowdown until concrete safety standards exist.

The practical obstacles are considerable, and the article's other facts point in that direction. Amodei himself wrote that some impactful forms of coordination are "legally challenging" and will require government support. Palantir CEO Alex Karp told CNBC that geopolitical adversaries make pausing almost impossible, and Chinese President Xi Jinping's Sunday BRICS remarks about a new AI open-source community and a digital ecosystem cloud platform suggest China has no interest in slowing down. Analyst Dion Hinchcliffe of SDA Bocconi offered four possible explanations for the three rivals' agreement — including that they saw something that "genuinely terrified them" — and cautioned against assuming the public one is the whole story. Whether any of this becomes an actual slowdown appears to hinge on whether government backing materializes and whether coordination can extend beyond these three companies.