Dario Amodei clears the air on banning open-weight AI models

July 28, 2026 - 12:31

Anthropic CEO Dario Amodei has pushed back against claims that he supports banning open-weight artificial intelligence models, clarifying that he never advocated for such a measure and does not believe it would improve safety. In a recent statement, Amodei said a ban on open-weight models would be ineffective and could even backfire by driving development underground or into less regulated jurisdictions.

The debate over open-weight AI models has intensified as companies release increasingly powerful systems that can be fine-tuned and deployed by third parties. Critics argue these models pose risks if misused for disinformation, cyberattacks, or autonomous weapons. Amodei acknowledged the concerns but stressed that outright prohibition is not the answer. Instead, he pointed to alternative approaches such as improved testing, transparency requirements, and voluntary safety commitments from developers.

Amodei's remarks come amid growing scrutiny of AI governance. Some policymakers have floated the idea of restricting open-weight releases, especially for frontier models that approach human-level capabilities in certain tasks. However, Amodei emphasized that collaboration between industry, researchers, and regulators is more productive than blanket bans. He also noted that open-weight models have enabled valuable research and innovation that could be stifled by overly aggressive restrictions.

The clarification from Amodei aims to reframe the conversation around practical safety measures rather than divisive calls for prohibition. As AI continues to evolve, the balance between openness and control remains a central challenge for the field.