Anthropic CEO Denies Open-Weight AI Ban After Industry Backlash

Anthropic PBC became the only major frontier AI lab to refuse signing an open letter titled "Open Weights and American AI Leadership," which was endorsed by over two dozen companies including OpenAI Inc., Google, Microsoft Corporation, Meta Platforms Inc., and Nvidia Corporation. The letter, led by Nvidia Corporation CEO Jensen Huang, urged U.S. policymakers to avoid premature restrictions on open-weight models—AI that developers can download and modify on their own infrastructure.

Dario Amodei, CEO of Anthropic PBC, published a blog post on July 28 denying that his company has ever advocated for a ban on open-weights models as a category. While Amodei characterized models without dangerous capabilities as a "public good" that expands access and strengthens competition, he disputed the industry letter's claims that open-weight models inherently favor cyber defenders over attackers or simplify the development of safeguards. He specifically cited risks that capable models could be used to weaponize pandemic-level viruses.

Amodei argued that protectionist bans would not address his primary national security concerns. Instead, he advocated for the Trump administration to focus on preventing authoritarian governments from accessing powerful AI chips, stopping industrial-scale distillation—a practice allegedly used by China's Moonshot to train its Kimi K3 model using Anthropic's Fable AI—and requiring safety testing for all capable models regardless of their weights' availability.