AI leaders want to hit the brakes after years of reckless speed

For years, the biggest frontier AI labs behaved like they were in a sprint to the finish line, with superintelligence or at least something close to it waiting there. This weekend, that tone started to break. The new message was less “go faster” and more “maybe don’t barrel straight into the wall.”

Anthropic’s Dario Amodei pushed hardest. In a nearly 4,000-word essay, he argued that the field “must slow the pace at which we improve the capabilities of AI models” because commercial pressure can drive a “race to the bottom” and make catastrophic risks worse. That is a notable shift for a sector that has spent years rewarding speed, scale, and bravado.

Then the rest of the leadership chorus joined in. OpenAI CEO Sam Altman said he agreed on social media and added that similar pacing talks had already been happening inside OpenAI. Alphabet chief scientist and Google DeepMind cofounder Demis Hassabis said Amodei was pointing “towards the right path forward,” and again called for an industry-wide standards body.

Microsoft CEO Satya Nadella also backed the mood change, saying the company welcomes the research, focus, and “deliberate pacing” needed to get alignment right. Microsoft is also preparing a lengthy “humanist AI” code of conduct for its models. The striking part isn’t that these companies are suddenly worried. It’s that several of the people pushing the field hardest are now saying the brakes should go on before the thing they built starts outrunning them.