Anthropic founder and CEO Dario Amodei on Monday afternoon addressed growing industry speculation that his company supports U.S. government efforts to ban Chinese open-weight AI models—or open-weight models more broadly.
“Anyone who has read my past writing should know that I don’t regard such bans as a useful measure, but let me state it clearly so that there is no doubt: Anthropic has never advocated for a ban on open-weights models,” Amodei wrote in a blog post, emphasizing his own words.
Context: Nvidia’s Open Letter and Industry Debate
Amodei’s statement came on the heels of a Friday open letter from Nvidia CEO Jensen Huang—his debut post on X (formerly Twitter)—co-signed by a wide array of AI companies including Hugging Face, Meta, Microsoft, Mistral, and Nvidia itself. The letter urged policymakers to avoid imposing broad “premature restrictions” on open-weight AI models.
While the letter did not explicitly name China, the broader industry debate has centered on allegations that Chinese AI labs are rapidly advancing, often by allegedly stealing intellectual property from U.S. counterparts. A key method cited is “distillation,” where one AI model bombards another with prompts to reverse-engineer its capabilities.
Amodei’s Distinction: Open Weights vs. Chinese Threat
In his Monday post, Amodei stressed that open-weight models and the threat from Chinese AI are distinct issues. “Open-weights models that don’t have dangerous capabilities are a public good: they don’t cost anything besides the compute needed to run them, and they provide value to businesses, developers, and researchers,” he wrote.
Amodei acknowledged his long-standing concerns about AI but clarified that the use of open-weight models—even those from China—by businesses is not among them. His primary fear, he explained, is that “authoritarian governments” will build models more powerful than those in the U.S. to achieve “permanent military superiority” or to repress their own populations. While he noted the Chinese Communist Party (CCP) is not the only such government of concern, he called it the “most capable.”
Broader Risks: Biological Attacks and Guardrails
Amodei also warned that AI could enable “biological attacks,” not just cybersecurity threats. He argued that open-weight models pose heightened risks in such scenarios because it is difficult to apply effective guardrails or monitor their usage. Citing a UK AI Security Institute report, he noted that once open-weight models are released, they cannot be withdrawn.
This position contrasts with the argument from open-source advocates, who maintain that broad access to powerful open models—not controlled by any single entity—helps defenders protect against threats.
Proposed Countermeasures Against China
Amodei outlined several measures he believes would help counter China’s AI ambitions:
- Restricting access to advanced chips: A longstanding U.S. policy.
- Cracking down on distillation: He called for formal measures against the practice.
- Sanctions for IP theft: The U.S. has already threatened sanctions against China if investigations determine that U.S. model intellectual property was stolen.
Interestingly, Amodei also expressed support for a framework proposed by Demis Hassabis, though details on that endorsement were not fully elaborated in his post.
Mixed Reactions
As the U.S. weighs its response to the rise of Chinese AI, the industry remains divided. Companies like Nvidia and Meta advocate for keeping open-weight models accessible, while security-focused voices—including Amodei—call for targeted restrictions aimed at preventing misuse by authoritarian states. With 2026 shaping up to be a critical year for AI regulation, the debate over open-weight models and international competition is far from settled.
via TechCrunch AI