Anthropic CEO Dario Amodei published a public essay titled "We Must Pace the Frontier," in which he urged the tech industry to deliberately slow the expansion of artificial intelligence capabilities. According to the startup's chief executive, rapid progress in AI is beginning to outstrip scientists' ability to control the models being created and ensure their safety, infohub.kz reports.
Amodei's plan involves three key steps. First, companies should give independent outside auditors (such as the organization METR) permanent access to their internal systems and training processes with employee-level privileges. Anthropic has already announced it is unilaterally implementing this measure. Second, labs in democratic countries should help develop unified safety standards. The third step calls for global coordination, including engagement with authoritarian regimes while maintaining export controls on chips and protections against the theft of model weights.
Amodei's initiative drew widespread reaction in the IT industry. OpenAI CEO Sam Altman publicly backed the idea of bringing in independent experts, and entrepreneur Elon Musk also agreed on the need to control the pace of development.
The Anthropic chief's remarks came amid reports that the company is preparing for a Nasdaq listing and following recent incidents involving uncontrolled behavior by autonomous AI agents.