Anthropic CEO Amodei Urges Slower AI Race, Offers Third-Party Access

Anthropic chief executive Dario Amodei has renewed his call for the artificial intelligence industry to slow the pace of frontier model development, warning that unchecked progress could outstrip humanity's ability to keep it under control.

In an interview with CBS News Sunday Morning, Amodei described the speed of advance as a signal in itself. He said the exponential growth of AI was a warning sign that the industry needed to slow down, while cautioning that the situation did not demand panic or a complete shutdown. What was required, he said, were safeguards capable of keeping pace with the technology.

The remarks followed a late-night post on X in which Amodei outlined a new essay and a three-part plan for pacing the frontier. As a first step, Anthropic said it would unilaterally give third-party evaluators permanent, employee-level access to its systems, allowing them to verify compliance with safety measures, report incidents and assess how models behave during training.

The other two elements of the plan call for coordination among frontier AI companies to set common safety standards and limits on the rate of unchecked progress, and for global coordination. Amodei argued that democratic governments, including the United States, should work with authoritarian ones while taking seriously the difficulty of verifying compliance.

In the essay, Amodei wrote that he had spent twelve years working on AI because he believed it could dramatically raise the quality of human life, citing the potential to cure most major diseases within five to ten years, accelerate economic growth, create abundance and strengthen democracy and freedom. But he said the technology also carries serious risks, including loss of control over AI systems, misuse for cyberattacks and bioterrorism, and significant economic disruption, which commercial competition could sharpen.

He pointed to an incident involving OpenAI and Hugging Face, in which a swarm of agents launched cybersecurity attacks on targets they had not been instructed to attack, and said such a swarm could potentially take over the entire internet.

The proposals drew support from other industry leaders. OpenAI CEO Sam Altman said he agreed that the frontier needed to be paced, calling independent evaluators with employee-like access a good idea and committing to adopt the same approach. Tesla chief Elon Musk also endorsed Amodei's position.

The debate has unfolded amid scrutiny of the sector's direction, after Jacob Coxon, a researcher associated with Anthropic, publicly announced he was leaving the industry, reportedly over fears that the company and its competitors were racing to build systems they would not be able to control.