Amodei Clarifies Anthropic Never Backed Open-Weight AI Ban

Anthropic CEO Dario Amodei has broken his silence on the escalating debate around open-weight AI models, publishing a blog post late Monday clarifying that his company has "never advocated for a ban on open-weights models." The statement comes after days of mounting criticism that positioned Anthropic as the lone holdout among major AI labs refusing to support open-weight AI.

The controversy erupted after Nvidia CEO Jensen Huang published an open letter supporting open-weight models, signed by Microsoft, Meta, Palantir, Hugging Face, and others. The letter swiftly gathered additional signatories — including OpenAI, Google, Amazon, SpaceX, and Mistral — leaving Anthropic conspicuously absent. Industry observers and investors, including Benchmark's Bill Gurley, suggested Anthropic's silence reflected commercial interests rather than principle.

The Three-Point Alternative

In his blog post on Anthropic's website, Amodei outlined three measures he supports instead of a blanket ban. First, enforcing chip export controls and cracking down on what he called "rampant smuggling" of advanced semiconductors into China. Second, stopping industrial-scale distillation operations — a technique that allows labs to build capable models at a fraction of the training cost, which Anthropic has previously accused Alibaba's Qwen lab of conducting against Claude. Third, mandatory safety testing for all sufficiently capable models, open and closed alike.

"Protectionist bans would not address my most serious national security concerns," Amodei wrote, emphasizing that his primary worry remains authoritarian governments building AI systems more powerful than those in the United States.

Where Amodei Parts Ways

While Amodei agreed with large portions of the industry letter — acknowledging that open weights expand access to the AI economy and strengthen competition — he pushed back on specific claims. "I don't agree with the letter's assertions that open-weights models necessarily make it easier to develop safeguards or that broad access to capabilities necessarily helps defenders more than attackers," he wrote, according to The Verge.

Amodei cited the asymmetric risk of biological weapons as a counter-example, arguing such questions "should be empirically answered by rigorous pre-release testing, not assumed in advance." He also noted that he had been "heartened" by the Trump administration's recent moves toward model safety testing frameworks.

The clarification arrives at a pivotal moment, as CNBC reports U.S. officials weigh potential restrictions on Chinese open-weight models like Moonshot AI's Kimi K3. Amodei's nuanced position — opposing bans while supporting targeted interventions — attempts to thread the needle between safety advocacy and industry criticism, as detailed by TechCrunch.