Anthropic’s Amodei: not backing an open-weight ban, but wary of authoritarian use and calls for global model testing
The stance influences how policymakers balance openness with security, potentially shaping export controls, safety standards, and international cooperation on AI governance.
At a glance
- Amodei said in a blog post that "Anthropic has never advocated for a ban on open-weights models."
- An open-letter led by Nvidia, and signed by companies including Hugging Face, Meta, Microsoft, and Mistral, urged policymakers not to impose premature restrictions on open-weight models.
- The UK AI Security Institute report cited in the discussion notes that once open-weights are released they cannot be withdrawn.
- Amodei advocates a global model safety testing organization with testing applied to the most capable models regardless of country of origin.
The story
TechCrunch reported on July 27, 2026 that Anthropic founder and CEO Dario Amodei responded to industry chatter that his company might support government bans on open-weight models, or on open-weights in general. He pushed back on the idea that Anthropic backs broad restrictions, stating clearly that his company has not advocated for a ban.