Anthropic publishes official position on open-weights models

Anthropic CEO Dario Amodei published a post on the company website clarifying the firm's position on open-weights models. Amodei said Anthropic has never advocated for a ban on open-weights models as a category and does not regard such bans as a useful measure. Amodei outlined two primary nightmare scenarios involving powerful AI in the hands of authoritarian regimes. He said one scenario involves authoritarian governments building superior AI models to achieve military dominance. The other involves powerful models being misused for cyberattacks or biological threats. Amodei said open weights are irrelevant to the first scenario and only a partial risk factor in the second. Amodei advocated for three strategic measures. He called for restricting powerful chips from going to China and cracking down on distillation operations, which he described as efforts from AI laboratories to extract the capabilities of a model to improve their own. He also called for implementing mandatory safety testing for all advanced models, regardless of whether they are open or closed. Amodei said Chinese interests have a chance to align on industry-wide safety testing and global cooperation, particularly in sectors such as bioweapon prevention.