Anthropic has announced the redeployment of its advanced AI cybersecurity model, Claude Mythos 5, to select U.S. organizations responsible for critical infrastructure. This decision follows a government-led review initiated on June 12, 2026.
Initially unveiled in April 2026, Claude Mythos 5 was described as a potential “cybersecurity reckoning” due to its unprecedented capabilities. The model autonomously identified thousands of high-severity software vulnerabilities across major operating systems and web browsers. Notably, it uncovered a 27-year-old flaw in OpenBSD and a 16-year-old vulnerability in FFmpeg, along with chained Linux kernel exploits enabling full privilege escalation. Impressively, Claude Mythos 5 achieved a 72% success rate in generating functional exploits and chaining vulnerabilities on the first attempt, a significant leap from the 0% success rate of its predecessor, the Opus model.
Given these powerful capabilities, Anthropic initiated Project Glasswing, a restricted-access program providing Claude Mythos 5 to approximately 200 vetted organizations, including U.S. government agencies and major tech companies like Apple, Amazon, and Microsoft. To support this initiative, Anthropic committed up to $100 million in Claude usage credits.
However, on June 12, 2026, Anthropic suspended access to both Claude Mythos 5 and Claude Fable 5 for all Glasswing partners. The company cited collaboration with the U.S. government to investigate the disruption and aimed to restore access promptly. This suspension temporarily deprived critical infrastructure defenders of the model’s advanced threat detection and vulnerability identification capabilities.
As of June 27, 2026, the U.S. government has authorized Anthropic to redeploy Claude Mythos 5 to a defined group of U.S. organizations operating and defending critical infrastructure. Anthropic is expediting the restoration of access for these validated entities and continues to work with the government toward broader availability of Mythos 5 and the return of Claude Fable 5, its most capable general-use model.
This phased restoration underscores the delicate balance between leveraging cutting-edge AI for national cybersecurity and managing the potential risks associated with such powerful tools. The collaboration between Anthropic and the U.S. government highlights the importance of controlled deployment to ensure that these advanced capabilities are used responsibly and effectively in safeguarding critical infrastructure.
