Anthropic CEO: We Never Advocated Banning Open-Weights Models

In a post dated July 27, 2026, Anthropic CEO Dario Amodei responded to ongoing debate around open-weights AI models — particularly those from China — and reports that some U.S. officials were considering banning their use by American companies. Amodei flatly stated: 'Anthropic has never advocated for a ban on open-weights models,' describing open-weights models without dangerous capabilities as 'a public good.' He outlined two key concerns: first, that authoritarian governments, particularly the CCP, could build AI more powerful than U.S. systems and use it for military superiority or mass repression — a risk he said is unrelated to whether weights are openly released; second, that powerful AI could be misused for cyberattacks or biological attacks. He acknowledged open-weights models present higher risk in the second scenario because guardrails are difficult to enforce, but argued banning their use by U.S. businesses would not address this since bad actors are unlikely to be legitimate U.S. companies. Instead, Amodei endorsed three measures: restricting chip and chipmaking equipment sales to China and cracking down on smuggling; targeting industrial-scale distillation operations that allow foreign actors to build capable models more efficiently than training from scratch; and unspecified additional measures referenced but cut off in the available text.