Anthropic CEO Dario Amodei calls for government power to block risky AI models

Dario Amodei, the CEO and co-founder of Anthropic, wants the US government to have the authority to stop AI developers from releasing models deemed too risky for public deployment.

The call isn’t new territory for Amodei, who has been beating this drum since he left OpenAI to found Anthropic back in 2021.

From op-eds to legislation

In June 2025, Amodei published an op-ed advocating for federal transparency laws and risk mitigation requirements for AI companies.

That advocacy has already found some legislative footing. California’s SB 53, enacted in late 2025, mandates safety testing and cybersecurity measures for frontier AI models. New York established its own set of AI regulations around the same time.

At the core of Amodei’s framework is what he calls “responsible scaling policies.” The concept is straightforward: the riskier the AI model, the more safety measures you need before you can deploy it. This framework has gained meaningful traction within the AI community, with other companies beginning to adopt similar tiered approaches to safety and deployment.

The Pentagon standoff

In February 2026, Anthropic refused Pentagon demands to modify its Claude models for military use. The military wanted Anthropic to lift certain safety safeguards.

The refusal triggered a temporary federal ban on Anthropic products, a designation as a supply-chain risk, and a cascade of legal disputes between the company and the US government.

The situation reached partial resolution in March 2026, when a federal court decision restored some access to Anthropic’s models.

One notable absence in Anthropic’s regulatory journey: there’s been zero intersection with crypto or digital assets. Despite the broader tech industry’s flirtation with blockchain-based AI projects and tokenized compute markets, Amodei’s safety-first framework operates in an entirely separate universe from the decentralized AI movement.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.