Anthropic gives thoughts on open-weights models, especially those from China
… Amodei outlines two primary nightmare scenarios: one where authoritarian governments build superior AI models to achieve military dominance, and another where powerful models are misused for cyberattacks or biological threats. …