Anthropic CEO Dario Amodei calls for slowing frontier AI progress

In short: Anthropic CEO Dario Amodei is urging AI companies to slow work on the most advanced AI models so safety work can keep up.

What happened

On September 12, 2026, Dario Amodei, the CEO of Anthropic, published an essay titled “We Must Pace the Frontier.” In it, he argued that companies should slow down how quickly they make “frontier” AI models more capable. Frontier models are the most advanced systems, like the fastest cars on the road.

Amodei did not call for stopping AI research completely. Instead, he described a slower, more coordinated approach, with clearer safety checks.

He suggested putting independent evaluators inside AI labs. These evaluators would get access similar to employees, so they can verify safety work and raise concerns. He also called for leading AI companies to agree on shared safety standards, and for governments to cooperate internationally on rules.

Some coverage has mixed this safety debate with a claim that Anthropic is expected to generate “$100 billion in annualized revenue” this year. Based on public reporting available as of mid-September 2026, that specific revenue figure does not match credible reporting about Anthropic’s finances. Numbers in the “hundreds of billions” that appear in related discussions typically refer to possible economic damage from AI misuse, not company revenue.

Why it matters

More powerful AI can be used for harmful things, like automated fraud or cyberattacks at large scale. Amodei’s argument is that if progress moves too fast, safety steps may become like adding seatbelts after the car is already speeding up. That could affect how AI tools are built, tested, and regulated in the coming years.

Source: NYTimes