Anthropic CEO Calls for Strengthening Frontline AI Regulation in the U.S.

Deep Tide TechFlow News. On June 11, according to Decrypt, Anthropic CEO Dario Amodei said in an article published on Wednesday that the U.S. government needs to impose binding safety requirements on the most powerful AI models, and that the era of transparency-based regulation has ended. In his article, “Policy and the Exponential Development of AI,” Amodei argues that frontier AI systems should undergo technical testing and audits like aircraft, and that any system that fails to meet safety standards should be barred from release.

Amodei’s proposal calls for mandatory third-party testing of advanced AI models, covering four categories of risks: cybersecurity, biological weapons, runaway risks, and automated R&D, and authorizes the government to prevent unsafe deployments. The proposal comes as Anthropic prepares for an IPO; it has previously completed a $6.5 billion Series H funding round with a valuation of $96.5 billion. Critics such as OpenAI CEO Sam Altman question whether stringent regulation could become a rationale for a small number of companies to consolidate control over the technology.