Anthropic CEO rejects open-weights ban, warns of authoritarian AI
Anthropic's Dario Amodei says chip bans, not model bans, are the real solution.
Anthropic CEO Dario Amodei clarified that the company has never advocated for banning open-weights models, calling them a public good when safe. He outlined two nightmare scenarios: authoritarian states building superior AI for military/repression, and misuse of advanced models for cyber/biological attacks. Instead of a blanket ban, he supports three measures: blocking chip sales and smuggling to China, cracking down on industrial-scale distillation, and requiring mandatory safety testing for all sufficiently capable models before release.
- Anthropic has never advocated for a ban on open-weights models; they are a public good when safe.
- Amodei's primary concern is authoritarian states building superior AI via secret training, not open-weights.
- He supports chip export restrictions and cracking down on distillation, not model bans, as effective measures.
Why It Matters
For tech professionals, this clarifies that AI policy debates should focus on hardware and distillation, not open-weights bans.