Anthropic CEO Dario Amodei urges slower rollout of top AI systems

In short: Anthropic CEO Dario Amodei is urging AI companies and governments to slow down and test the most advanced AI systems more carefully.

What's going on

Dario Amodei, the CEO and co-founder of Anthropic, has become a leading voice arguing that powerful AI could cause serious harm if it improves faster than society can understand and control it. He also says AI could bring big upsides, like faster progress in science and medicine.

In a September 2026 essay, Amodei called on the industry to “pace the frontier.” In plain terms, he wants tougher safety tests before releasing the most capable AI models, and ongoing access for independent evaluators (outside experts who can inspect and test systems, like auditors checking a company’s books). He also warned that AI is getting better at helping build even more capable AI, which could speed progress and raise risks.

His stance has created tension with some peers. Meta CEO Mark Zuckerberg and Nvidia CEO Jensen Huang have described him as a “doomer,” and some critics say strong public warnings could also help Anthropic argue for tighter controls that may favor its position. Separately, Anthropic has clashed with the US government, including over refusing to allow its AI model Claude to be used for domestic mass surveillance or fully autonomous weapons.

What to watch

The debate is no longer just about ideas, it is about rules and access. Some rivals, including Sam Altman and Elon Musk, have publicly agreed with the general idea of slowing down, but business and political leaders remain split. Watch for whether governments require independent testing and ongoing monitoring for the most advanced AI, similar to safety checks used in other high-risk industries.

Source: NYTimes