Our position on open-weights models

Anthropic CEO Dario Amodei clarified that the company does not support banning open-weights models, despite recent speculation, and argues that protectionist bans would not address genuine national security concerns. His primary worry is that authoritarian governments could develop more powerful AI systems to achieve military superiority or enable mass repression, regardless of whether models are open-source. While acknowledging that open-weights models do present alignment and misuse risks, he contends that banning their use by US companies would only protect domestic AI companies from competition rather than meaningfully address security threats.