Anthropic CEO outlines plan to slow AI development

Anthropic CEO Lays Out Plan to 'Pace the Frontier' of AI Development

Amid escalating warnings from AI researchers about existential risks, Anthropic CEO Dario Amodei has published a detailed plan to deliberately slow the development of advanced AI capabilities. The proposal comes shortly after a researcher resigned from Anthropic citing concerns the company was “gambling with our lives.” Significantly, the initiative has already received public support from OpenAI CEO Sam Altman and Elon Musk, signaling a potential shift in strategy among leading AI labs from unrestrained progress to coordinated caution.

Amodei's post outlines a three-part strategy for managing the pace of AI advancement, prompted by the technology's rapid recent progress and its growing ability to help build the next generation of AI. Anthropic is unilaterally committing to the first step and is calling on other labs and governments to follow suit. The core proposals include:

• Embedded Evaluators: Inviting third-party auditors, with access comparable to internal risk teams, to verify safety commitments and report incidents. Anthropic is implementing this immediately, and OpenAI has publicly agreed to do the same.
• Coordinated Safety Standards: Calling for leading AI companies in democratic countries to establish common safety protocols and limits on unchecked progress, with the US government mediating discussions to navigate potential antitrust concerns.
• Global Coordination: Attempting to cooperate with authoritarian governments, including China, to prohibit specific dangerous AI applications, such as the development of biological weapons.

This public call for a slowdown positions Anthropic and OpenAI at the center of the AI safety debate, moving the conversation from abstract fears to concrete policy proposals. While supporters see it as a responsible step, the plan has drawn skepticism from critics who argue it could be a form of regulatory capture. Such proposals, they suggest, could entrench the market dominance of established players by creating high compliance barriers for new competitors, while also distracting from the immediate, real-world harms AI is already causing.

While framed as a crucial safety measure, this coordinated push for 'pacing' development led by Anthropic and OpenAI is also a strategic move to shape future regulation, potentially cementing their market leadership by defining the terms of compliance for the entire industry.