Anthropic Clarifies Position Against Open-Weights Model Bans

Anthropic reiterated clearly that it does not advocate for bans on open-weights models. The company states that open-weights models without dangerous capabilities are a public good. Anthropic agrees that open-weights models expand access to the AI economy and strengthen competition, but disagrees that they necessarily facilitate development of safety protections. The company proposes three measures: legal and commercial frameworks focused on distillation countermeasures, robust protections against dangerous capability access, and rigorous independent safety evaluations. Anthropic maintains concerns about broad access to dangerous capabilities and attacker-defender asymmetry in domains like biology.