Anthropic CEO Calls for Slowing Frontier AI, Lays Out Three-Step Plan

Anthropic chief executive Dario Amodei said the time has come to slow down AI development. He laid out the argument in a proposal published under the title "We Must Pace the Frontier" Dario Amodei, reported on Sept. 12, 2026 The Verge.

The proposal centers on a three-step plan called "pace the frontier" to slow the pace of AI training and development. It sequences unilateral action by Anthropic, then industry-wide coordination, then international agreement. The plan is framed as a proposal, not as a policy already in force.

The first step is unilateral. Anthropic will give third-party evaluators such as METR access to its models to help ensure adherence to safety practices and commitments. Under the plan, those embedded evaluators will have employee-like access Dario Amodei. Anthropic is taking that step now, on its own.

The second step moves beyond one lab. Amodei calls for industry, likely with government agencies, to establish common safety standards and limits on the rate of unchecked AI progress. The mechanism is coordination on evals, thresholds, and training governance, rather than isolated company policies.

The third step is geopolitical. Amodei calls for getting authoritarian governments like China and Russia to agree to slow development and adopt global AI safety standards. That would extend the same logic, common standards and verifiable restraint, from industry to states.

The language echoes prior Anthropic positioning. The company has stated that it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment. In June, Anthropic called on major AI labs to consider a coordinated and verifiable pause in AI development if risks rise Reuters. On Aug. 31, Anthropic resumed external cyber testing of its AI models after security incidents involving Claude AI hacks Reuters.

For practitioners, the near-term detail that matters is evaluator access. Pre-deployment evals, red-teaming, and capability assessments already shape release decisions at frontier labs. Employee-like access for an external party changes that workflow. It implies persistent visibility into checkpoints, scaffolding, and system prompts, not just API sampling of a finished model. It also raises operational questions around containment, data handling, and disclosure timelines that labs and evaluators will have to negotiate in contracts and security reviews.

The broader context here is worth flagging, and in this author's view it is where the proposal will be tested. Voluntary evaluator access is implementable by one company. Common industry limits require shared definitions of what counts as frontier, what triggers a slowdown, and who verifies compliance. International restraint adds export controls, compute monitoring, and inspection to that list. Each layer is harder than the last, and the technical community will recognize the pattern from other dual-use regimes where measurement came before enforcement.

Looking at what this means for builders, the practical shift would be slower iteration at the top end in exchange for more legible safety work. That does not stop product development on existing capabilities. It does put more weight on inference optimization, eval harnesses, interpretability tooling, and deployment controls. Over the long arc, that kind of infrastructure has tended to compound. Better evals and clearer standards make it easier, not harder, to ship systems that enterprises and users can trust.