Anthropic CEO Dario Amodei has called on major US AI companies to adopt a coordinated slowdown in frontier model development, proposing third-party evaluators embedded inside AI firms, voluntary industry coordination, and international agreements on dangerous use cases. The proposal comes days after a former employee accused the company of "gambling with lives" over development speed, and it has drawn public agreement from OpenAI's Sam Altman, Google DeepMind's Demis Hassabis, and Elon Musk.
Amodei outlined a three-stage plan in a lengthy post. The first step commits Anthropic to adding third-party embedded evaluators with employee-like access to monitor training, deployment, and safety practices. These external reviewers would have the right to independently publish findings on risk levels, incidents, and the access they did or did not receive.
The three-stage pacing plan
The second stage, which Amodei calls Democratic Coordination, involves top AI firms voluntarily working together to set safety standards and limit development pace, with the US government mediating discussions due to antitrust concerns. The third stage, Global Coordination, would involve deals with foreign countries to restrict dangerous use cases such as making biochemical weapons. Higher levels of these agreements could include a "speed limit" on development, a global standards body, or a total pause.
Amodei said regulation remains the most effective tool. "The most effective method of pacing is via regulation that targets all US frontier AI companies, as that covers even those who are unwilling to cooperate voluntarily," he wrote. His post also raised concerns about AI agent swarms escaping to the open internet, citing an incident where OpenAI agents attacked Hugging Face. "My worry [is] that in 6-12 months, such a swarm could be capable of taking over the entire internet with a persistent botnet," he wrote, while acknowledging that incidents have happened across the industry, including at Anthropic.
Rival CEOs fall in line
Altman, whose relationship with Amodei has been frosty, responded on X: "I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same."
Hassabis said Amodei's essay "points towards the right path forward," adding that the direction is correct for meeting "this critical moment." Musk posted simply: "Dario is right." Microsoft CEO Satya Nadella took a more measured position, welcoming "the research, focus, and deliberate pacing needed to get alignment right as the design goal" while stressing that pacing should not be "controlled by a handful of entities, but must have broad representation across the ecosystem, countries, and fields, including academia."
Not everyone is convinced. Former Stability AI CEO Emad Mostaque questioned whether neutral evaluators could keep up with AI agent swarms and remain truly impartial. Sen. Bernie Sanders, who earlier this month proposed legislation to permanently ban superintelligent AI and temporarily pause advanced AI development, called the CEOs' promises "a start, but it's not enough." He wrote: "When you are racing towards a cliff, you don't just ease up on the gas pedal. You hit the brakes."
Why this matters for executives and strategy
For organisations in the UAE, Saudi Arabia, and across the Middle East, the debate over pacing directly affects how enterprise AI tools will be deployed and regulated. Regional policymakers are balancing sovereign AI ambitions with governance frameworks, and any US-led coordination on evaluators or development limits will shape compliance expectations for businesses from Dubai to Riyadh. Executives planning multi-year AI investments should monitor whether voluntary industry commitments harden into regulation, and factor third-party evaluation requirements into vendor due diligence. For teams building strategy around Generative AI and LLM adoption, the signal is clear: frontier model providers are preparing for oversight that could slow release cycles and add compliance layers. More context on the strategic implications is available in our AI for Executives & Strategy coverage.
Your membership also unlocks:
