Wire · regulatory
Amodei: No Open-Weights Ban, But Test Any Capable Model
◆ Sectors
◆ Geography
◆ Source
◆ Verified
Fusion42 · 28 July 2026 · Fusion42 review
Anthropic's Dario Amodei published a position paper rejecting claims the company sought to ban open-weights models, instead proposing three capability-agnostic measures: US chip export controls to China, industrial-scale distillation policy, and mandatory pre-release safety testing for any sufficiently capable model regardless of weights access. The move comes after Anthropic notably declined to sign Nvidia's open-weights coalition letter alongside other major AI labs, positioning the company as isolated in the AI safety debate.
This Wire brief sits within Fusion42's coverage of AI Frontier Models, and 2 sources have reported it between 27 Jul 2026 and 29 Jul 2026.
◆ ◆ The Wire takeaway
Anthropic just reframed the open-weights fight as a false choice: the real policy battle is over who gets tested before release and whether distillation becomes a regulated industrial process. If you're training models at scale, the threshold for mandatory safety testing is now the live policy question—it will either stay theoretical or become statutory.
◆ Coverage
2 sources · first reported 27 Jul 2026 · latest 29 Jul 2026
◆ Related on Wire
◆ Topics