Anthropic's Amodei says company never sought an open weights ban, but wants chip controls and mandatory safety tests

7:00 AM · July 28, 2026

Anthropic published a blog post on July 27, 2026 authored by CEO Dario Amodei laying out, for the first time in detail, the company's position on open weight AI models, an unusually public break from the industry consensus forming around Nvidia's new security alliance and an earlier letter signed by Nvidia, Microsoft, Meta, Palantir and more than 20 other firms. “Anthropic has never advocated for a ban on open weights models,” Amodei wrote, adding that “open weights models that don't have dangerous capabilities are a public good.” He argued that “a blanket ban on open weights models is neither the correct remedy” for national security concerns, since “protectionist bans would not address my most serious national security concerns.” Instead, Amodei laid out three preferred interventions: restricting sales of advanced semiconductors and chipmaking equipment to China and cracking down on smuggling, enforcing against industrial scale ‘distillation,’ where one company extracts a rival's model capabilities cheaply, and requiring mandatory pre release safety evaluations covering cyber, biological and alignment risks for all sufficiently capable models regardless of whether they are open or closed. The post explains why Anthropic did not join the recent open weights letter or Nvidia's new alliance: Amodei said he rejects the assumption that wider model access necessarily helps cyber defenders more than attackers.