Labs

Anthropic's Fable 5
stays dark on
day nine as
Trump softens, Commerce
directive holds

Nine days into an unprecedented export-control suspension of Claude Fable 5 and Mythos 5, the Commerce Department's order is still in force — even as 150-plus executives demand it be lifted and the administration's own rationale, sourced to a conversation with Amazon's Andy Jassy, comes into focus.

By Casey Vanderhoof Frontier labs correspondent · June 21, 2026

Nine days after the Commerce Department’s 5:21 p.m. ET directive of Friday, June 12, Anthropic’s two most capable models, Claude Fable 5 and Mythos 5, remain disabled for every customer worldwide. The order, which bars access by any foreign national including Anthropic’s own non-citizen US-based employees, left the company no operational path short of full shutdown. Fable 5 had been live for three days.

The stated rationale was national security. The actual mechanism, now visible thanks to Wall Street Journal reporting surfaced by CNBC on June 17, looks closer to competitor relay. Amazon CEO Andy Jassy reportedly told Treasury Secretary Scott Bessent and other administration officials that Amazon researchers had coaxed Fable 5, via a sequence of prompts, into producing information useful for cyberattacks. Bloomberg separately confirmed Commerce as the originating agency. Anthropic, after reviewing a demonstration of the technique, says it amounted to prompting the model to read a codebase and surface software flaws, a capability already available in OpenAI’s GPT-5.5.

What’s striking is the regulatory architecture, or absence of it. There’s no published rule, no formal finding, no appeal process visible from outside the building. “This sure looks mandatory, if there are going to be consequences for not doing what the government says,” said Daniel Remler of the Center for a New American Security. Alex Stamos, chief product officer at Corridor and Facebook’s former security chief, organized an open letter to Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross, signed by more than 150 executives and technical leaders, demanding the directive be lifted. “Those rules need to be written down and transparent,” Stamos said. “That has not happened.”

The political backdrop makes the absence of process legible. In February, Trump ordered federal agencies to stop using Anthropic’s models after the company refused the Pentagon’s preferred contract terms. In March, the Defense Department designated Anthropic a “supply chain risk,” a label the company is challenging in federal court. Former AI czar David Sacks has publicly called Anthropic “woke” and accused it of regulatory capture.

And then the irony, which is doing more work in this story than any single fact. Days before the directive landed, CEO Dario Amodei published an essay arguing that frontier models should face mandatory technical testing and that unsafe releases should be “blocked or reversed.” The Commerce action is, structurally, exactly that, minus every statutory guardrail Amodei was asking for. The closest precedent isn’t an export-control case at all; it’s the post-9/11 National Security Letter regime, where the substantive authority was real and the procedural transparency wasn’t.

Anthropic asked for a regime with teeth. It got the teeth.

Sources