Anthropic has issued a statement clarifying that it has never advocated for a ban on open‑weights AI models, a position that has come under scrutiny amid discussions about potential restrictions on Chinese‑origin models. The company argues that open‑weights models that lack dangerous capabilities represent a public good, providing value to businesses, developers, and researchers without significant cost beyond compute. Anthropic notes that protectionist bans would not address its most serious national‑security concerns, citing two nightmare scenarios outlined in a prior essay. To address these concerns, Anthropic supports three measures: keeping powerful chips out of authoritarian hands, stopping industrial‑scale distillation, and requiring safety testing for all sufficiently capable models, whether open or closed. The company also highlights ongoing safety evaluations with METR, early discussions with Hugging Face, and its recent expansion into Singapore, where it signed an MOU to promote responsible AI adoption. These actions underscore Anthropic’s commitment to balancing openness with rigorous safety protocols in the evolving AI landscape.
Our position on open-weights models
Anthropic clarifies it has never supported a ban on open‑weights AI models, arguing that such models are a public good and that protectionist bans would not address national security concerns. The company stresses the need for safety testing, limiting powerful chips to non‑authoritarian actors, and preventing industrial‑scale distillation. Anthropic also notes ongoing collaborations with METR, Hugging Face, and a Singapore office to advance safe AI.
Why it matters
The stance influences policy debates on AI openness and security, shaping how companies balance innovation with risk mitigation.