Anthropic CEO clarifies stance on open-weights AI models

Anthropic CEO Dario Amodei published a post clarifying the company's position on open-weights AI models, denying that Anthropic has ever advocated for a ban on such models. He distinguishes between two nightmare scenarios: authoritarian states (especially China) building more powerful AI for military or surveillance purposes, and misuse of powerful models for cyber or biological attacks. Amodei argues that banning open-weights models used by US businesses does not address either threat. Instead, he supports three measures: restricting chip sales to China, cracking down on industrial-scale distillation, and requiring mandatory safety testing for all sufficiently capable models, open or closed. He notes that the Trump administration has moved toward such testing, and that industry proposals exist to apply testing regardless of origin or openness. Amodei agrees with an open letter supporting open weights for expanding access and competition, but disagrees that open weights necessarily aid defenders more than attackers, citing potential biological risks. He emphasizes that safety questions should be answered by empirical testing, not assumptions.