Anthropic CEO Dario Amodei published an essay titled We Must Pace the Frontier on September 12, 2026. The document proposes a three-step framework to balance the speed of AI development with safety research, following an open letter signed by over 1,100 AI professionals in July 2026.
The first step involves implementing embedded evaluators immediately. These tools will assess and verify safety practices during model training and deployment, replacing one-time audits with a continuous feedback loop. This is the only component Anthropic intends to execute unilaterally.
The second step, democratic coordination, focuses on aligning major AI labs and democratic governments on shared safety standards. This aims to prevent companies from gaining competitive advantages by neglecting safety protocols. The third step, global coordination, seeks to establish international norms for AI development.
This framework builds upon Anthropic’s existing Responsible Scaling Policy, which utilizes AI Safety Levels to trigger safeguards based on capability thresholds. Amodei, who co-founded Anthropic in 2021, previously authored essays on AI discipline, including Machines of Loving Grace in 2024 and The Adolescence of Technology in January 2026.
