Anthropic CEO Dario Amodei rejected calls for a ban on open-weight AI models, labeling safe versions a public good. This statement follows criticism of Anthropic for refusing to sign an industry letter supporting open-weight AI. Competitors including Google, OpenAI, and Meta endorsed that letter.
Amodei warned that authoritarian regimes like China could exploit powerful AI for military superiority or domestic repression. He cautioned that advanced open-weight models risk misuse for cyber or biological attacks because safeguards are easily removed.
The CEO proposed mandatory safety testing for all highly capable models regardless of their release format. He also advocated for stricter export controls on advanced semiconductors to mitigate these risks.