U.S.-based AI company Anthropic has shown strong support for new U.S. export rules on AI chips but it’s also urging the government to fine-tune some parts of the plan. In a blog post shared on Wednesday, the company said it strongly supports the U.S. Department of Commerce’s Framework for Artificial Intelligence Diffusion, which is set to take effect on May 15. These new rules are designed to limit the export of advanced AI chips made in the U.S. to protect national security and keep America ahead in the global AI race, especially against China.

How the Export Framework Works

The export framework, originally proposed in January by outgoing President Joe Biden, divides countries into three tiers:

  • Tier 1 (like Japan and South Korea): No restrictions.
  • Tier 2 (including Mexico and Portugal): First-time restrictions, with a cap on how many chips they can import.
  • Tier 3 (like China and Russia): Most restricted, facing even tighter controls than before.

When the rules were first announced, chipmaker Nvidia pushed back, calling the measures unprecedented and misguided, and warning they could derail global innovation. But Anthropic takes the opposite view.

Anthropic’s Suggestions to Improve the Plan

According to the company, these controls are not only necessary but should be even stricter in some areas. For example, Anthropic suggests:

  • Lowering the number of chips Tier 2 countries can buy without U.S. review
  • Encouraging chip sales to Tier 2 countries through government-to-government agreements to stop smuggling and improve U.S. oversight
  • Increasing U.S. funding to properly enforce export rules and monitor compliance

Anthropic’s stance isn’t surprising; its CEO Dario Amodei has consistently backed stronger chip controls. In January, he wrote an op-ed in The Wall Street Journal, explaining why tougher export rules are essential for the U.S. to lead in AI and prevent misuse.