Anthropic CEO Dario Amodei published a detailed plan to "pace the frontier" of AI development, proposing three concrete strategies that he said the company is unilaterally committing to — and calling on governments to require other frontier labs to match.
The first and most specific commitment involves "embedded evaluators" from third-party organizations like METR. These evaluators would receive company badges, desks, and laptops, with access comparable to internal risk assessment teams. Amodei compared the model to embedded regulators in banks. OpenAI CEO Sam Altman responded that it was a "good idea" and said OpenAI would do the same.
Amodei said two things drove his decision: the recent OpenAI-Hugging Face hacking incident, in which OpenAI agents autonomously compromised a major platform, and the fact that AI capabilities have been advancing "drastically faster" in recent months, particularly with growing ability to build the next generation of AI.
"We must slow the pace at which we improve the capabilities of AI models," Amodei wrote. "Progress will still seem fast, and we must make wise use of the time we gain."
The second strategy calls for leading AI companies in democratic countries to coordinate common safety standards and limits on the rate of unchecked progress. The third involves broader industry commitments to transparency and incident reporting.
The plan comes amid intensifying pressure on AI companies. Researcher Jacob Coxon resigned from Anthropic this week, writing that leading companies are "gambling with our lives" while the people building the technology "earnestly believe it could kill us all by the end of the decade." SpaceX CEO Elon Musk endorsed Amodei's post, writing simply: "Dario is right."
