Amodei: Anthropic never pushed to ban open weights, it wants pre-release testing

After a week of criticism, Dario Amodei published a letter saying “Anthropic has never advocated for banning open weights.” Models without dangerous capabilities are, in his words, a “public good” that costs nothing beyond the compute to run them.

His proposal has three parts: tighten export controls on advanced chips and the equipment used to make them, shut down industrial-scale distillation, and require a mandatory safety review of every “sufficiently capable” model before release. The letter doesn’t say where that threshold sits or who runs the tests, The New Stack notes.

The rival statement, “Open Weights and American AI Leadership,” has 50 signatures. OpenAI and Google joined Microsoft, Nvidia, Meta, AMD, GitHub and Ollama on it. Anthropic and Amazon did not sign.