Anthropic Confirms Claude Mythos 5 Redeployment for US Critical Infrastructure Organizations
June 27, 2026 — Anthropic confirmed the U.S. government authorized redeployment of Claude Mythos 5 to U.S. organizations operating and defending critical infrastructure, following a review process that began with the June 12, 2026 suspension of Mythos 5 and Fable 5 for all Project Glasswing partners.
Background
Claude Mythos first gained attention in April 2026 when Anthropic described potential “cybersecurity reckoning” capabilities — autonomously discovering vulnerabilities including a 27-year-old OpenBSD flaw and 16-year-old FFmpeg flaw. The model achieved ~72% first-attempt exploit chaining success vs 0% from prior Opus.
Project Glasswing restricted access to ~200 vetted organizations (U.S. government agencies, Apple, Amazon, Microsoft, etc.) with up to $100M in Claude usage credits committed.
June 12 Suspension
Anthropic suspended Mythos 5 and Fable 5 for all Glasswing partners on June 12, 2026, citing coordination with the U.S. government. Critical infrastructure defenders lost access to advanced threat detection during the review window.
June 27 Partial Restoration
The U.S. government notified Anthropic that Mythos 5 can redeploy to organizations operating/defending critical infrastructure — energy, healthcare, financial services, telecommunications, and similar sectors.
Reports indicate 100+ organizations may receive access including Fortune 500 companies, government agencies, and Glasswing participants. Approved organizations may no longer need export licenses for foreign-national employees in some cases.
Claude Fable 5 remains suspended for general use. Anthropic continues working toward broader Mythos 5 access and Fable 5 restoration.
Anthropic statement (June 27, 2026): “Since June 12, we’ve been working closely with the US government to restore access to Claude Mythos 5 and Fable 5. Today, the government notified us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical infrastructure.”
Implications
Phased restoration signals cautious U.S. government approach to harnessing frontier AI for national cyber defense while controlling offensive dual-use risks. Follows OpenAI’s June 26 GPT-5.6 limited preview under similar government coordination.