Anthropic: we never advocated banning open-weights models

On 27 July 2026, Anthropic CEO Dario Amodei published a post on the company's newsroom responding to the debate of the past few days around open-weights models, particularly those from China. The context: reports suggest some US officials are considering banning US companies from using Chinese open-weights models; many tech companies signed a letter supporting open weights; and some have accused Anthropic of wanting a ban to protect its own business.

Amodei states it plainly: Anthropic has never advocated for a ban on open-weights models. In his view, open-weights models that don't carry dangerous capabilities are a public good — they cost nothing beyond the compute to run them, and they provide value to businesses, developers and researchers.

Key points

• Amodei's two nightmare scenarios: (1) authoritarian governments building models more powerful than the US to achieve military superiority or deep domestic repression — open vs. closed is irrelevant here; (2) powerful models misused for cyber or biological attacks, plus serious alignment problems.
• Why a ban doesn't fix it: the most dangerous model might be one trained in secret and never released. And banning US business use does nothing about bad actors, who are unlikely to be legitimate US businesses.
• The three measures Anthropic does support: no sales of powerful chips or chipmaking equipment to China (plus a crackdown on smuggling); action against industrial-scale distillation; mandatory pre-release safety testing for ALL sufficiently capable models, open and closed.
• Where he agrees with the open letter: open weights expand access to the AI economy, strengthen competition for at least some use cases, and give customers greater control.
• Where he disagrees: he rejects the claim that open weights necessarily make safeguards easier to build, or that broad access to capabilities necessarily helps defenders more than attackers — he thinks the opposite is at least as likely, and that testing should answer this empirically.
• The irreversibility problem: once weights are released they cannot be withdrawn, and it is very difficult to apply guardrails or monitor usage.

FAQ

Has Anthropic asked for a ban on open-weights models? No. The post states directly that Anthropic has never advocated for a ban on open-weights models as a category.

What does Anthropic want restricted instead? Three specific things: exports of powerful chips to China, industrial-scale distillation operations, and the release of sufficiently capable models without safety testing — applied to open and closed models alike.

What does this mean for developers using open models? The post frames open-weights models without dangerous capabilities as a public good that benefits developers. The thing to watch is the direction of pre-release testing policy, since that will shape which models ship and when.