- Anthropic says open-weight models without dangerous capabilities are a public good and should not face a blanket ban. [1] - Amodei supports tighter chip controls, action against industrial-scale distillation and mandatory testing for sufficiently capable open and closed models. [1] - Anthropic declined to sign an industry letter because it disputes the signatories’ claims about the security benefits of broad model access.
[1][2] Anthropic CEO Dario Amodei said Monday that his company has never advocated banning open-weight artificial intelligence models, responding to criticism after Anthropic declined to join a broad industry statement defending downloadable models.[1]
Amodei’s position leaves Anthropic aligned with open-model advocates on competition and access, but divided from them over safety. He called open-weight models without dangerous capabilities “a public good,” while arguing that powerful downloadable systems may be harder to monitor, safeguard or withdraw after release.[1]
Where Anthropic breaks with the industry letter #
The July 24 letter, signed by companies including Nvidia, Microsoft, Meta, Hugging Face, Google and OpenAI, urged policymakers to avoid premature restrictions. It argued that open weights expand access, reduce dependence on individual vendors and let more researchers examine models and develop safeguards. It also called for targeted legal remedies against unlawful extraction from proprietary systems rather than sweeping restrictions on distillation techniques.[2]
Amodei said he agrees that open weights can strengthen competition and give customers more control. Anthropic did not sign because he rejects the assumption that wider access necessarily helps defenders more than attackers or makes safeguards easier to build. He called for empirical, pre-release testing instead.[1]
His preferred policies are restrictions on advanced chip and chipmaking-equipment sales to China, enforcement against industrial-scale distillation and mandatory safety evaluations for all sufficiently capable models, regardless of whether their weights are released.[1]
Open-model advocates remain more confident that transparency improves security. Amanda Brock, CEO of OpenUK, said risks are better managed “openly, transparently and collaboratively,” endorsing the security rationale behind the industry letter. [3] Anthropic’s statement narrows the dispute: the company opposes a categorical ban, but it has not accepted the open-model coalition’s broader safety case.
Companies mentioned #
Further sources #
The stories that matter, in one email. Free — unsubscribe anytime.