Anthropic’s Amodei rejects claim it wants to ban open-weight AI models

11:00 AM ET · July 30, 2026

Anthropic chief executive Dario Amodei used a blog post this week to push back on criticism that his company is quietly campaigning to ban open weight AI models, after Anthropic became the only major frontier lab to decline signing a letter, organized by Nvidia chief executive Jensen Huang, urging US policymakers to avoid premature restrictions on open weight systems. More than fifty companies, including Microsoft, Meta, Google, OpenAI and Hugging Face, ultimately signed that letter. Amodei wrote that Anthropic has “never advocated for a ban on open-weights models” as a category, and described open models that lack dangerous capabilities as “a public good.” His stated concern is narrower: that authoritarian governments could use open, downloadable models more capable than anything available today to pursue military advantage or domestic repression, systems that, once released, are difficult to monitor, safeguard or withdraw. He pointed to warnings from Vice President Vance and the US intelligence community’s 2026 Annual Threat Assessment about rival powers closing the AI capability gap as evidence the concern is shared inside government. The episode has become a visible fault line among AI labs, with Anthropic now publicly isolated on an issue where its two largest closed model rivals, OpenAI and Google, both signed onto the industry consensus over the same weekend.