Anthropic Pulls Fable 5 Offline After U.S. Export Order
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic received the export-control directive at 5:21 p.m. ET Friday and said it will "abruptly disable" Fable 5 and Mythos 5 for all users, not just foreign nationals, while it works to restore access.
- The U.S. government cited a potential "jailbreaking" method that lets Fable 5 identify software vulnerabilities, but Anthropic said the technique is narrow and that similar capabilities exist in other publicly-available models, naming OpenAI's GPT-5.5 specifically.
- Fable 5 and Mythos 5 launched days earlier; Mythos 5 has safeguards lifted in areas like cybersecurity and remains accessible only to a vetted group of cyber defenders and critical infrastructure operators.
- Anthropic's Red Team reported that Mythos-class models can convert newly disclosed software vulnerabilities into working exploits in hours or minutes, turning "N-days into N-hours" — a capability the company framed as a strategic advantage for attackers.
- The Department of Defense earlier this year labeled Anthropic a "supply chain risk" after the company sought to draw red lines on military use of its technology; Anthropic has filed two lawsuits challenging that designation.
Why it matters: The government moved on a single reported jailbreak technique that Anthropic itself says other frontier models can replicate, making the ban a precedent-setting test of whether export controls will follow the capability — or the company. With Anthropic already in litigation over a DoD "supply chain risk" label, the order deepens its standoff with Washington and hands European policymakers (per the AP-sourced reaction in other coverage) a fresh argument for sovereign AI investment.
