Bottom line: Anthropic rejects bans on open-weights models and instead proposes technological measures such as chip sanctions against China and control of distillation operations.
Anthropic CEO Dario Amodei pushes back against allegations that the company wants to ban open-weights models. Instead, he advocates for chip export restrictions and limitations on model distillation to address security risks.
Dario Amodei, CEO of Anthropic, positions himself clearly against speculation that his company is seeking a ban on open-weights models. In a statement, he emphasises: “Anthropic has never advocated for a ban on open-weights models.” This clarification comes against the backdrop of reports that US authorities are considering restricting the use of Chinese open-weights models by American companies, and after several tech companies signed a letter in support of such models.
Amodei defines his two main concerns: First, the worry that authoritarian governments – particularly China – will develop AI models that surpass US systems, leading to military superiority or domestic repression. Second, the risk that powerful open models could be misused for cyberattacks or biological attacks, as security measures are difficult to enforce in open-weights models and once released, weights cannot be retrieved. For CTOs, it is relevant to note: according to Amodei, open-weights models in themselves are not a problem if they do not possess dangerous capabilities.
In response to these security concerns, Amodei advocates for three concrete measures: (1) export restrictions on high-performance chips and chip manufacturing to China, since China cannot develop superior models without US technology due to scaling laws; (2) measures against large-scale model distillation, a computationally efficient process through which China could partially circumvent chip shortage constraints; (3) further measures that Amodei only partially details in his statement. These measures, according to Amodei, address security risks more directly than industry-wide bans.
Source: www.anthropic.com · Published 26 July 2026
Lumi AI News — AI-assisted curation in accordance with Art. 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.3.