“`html
Anthropic has affirmed that Claude Mythos 5, its most potent AI cybersecurity model, will be reintroduced to a select group of U.S. organizations tasked with operating and safeguarding essential infrastructure, following a government-supervised review process that commenced on June 12, 2026.
Claude Mythos first garnered attention in April 2026 when Anthropic characterized it as a potential “cybersecurity reckoning,” an AI model so proficient that the firm initially opted not to publicly disclose it.
The model showcased an unparalleled capacity to independently uncover software weaknesses, revealing thousands of high-severity vulnerabilities across various major operating systems and web browsers.
Among its reported discoveries were a 27-year-old flaw in OpenBSD and a 16-year-old vulnerability in FFmpeg, alongside linked Linux kernel exploits capable of full privilege elevation.
The model reportedly attained a 72% success rate in generating operational exploits and chaining vulnerabilities on the initial attempt — in contrast to 0% from the earlier Opus model.
Given these competencies, Anthropic initiated Project Glasswing, a closed-access program that limited access to around 200 approved organizations, including U.S. government entities and major tech partners such as Apple, Amazon, and Microsoft. Anthropic pledged up to $100 million in Claude usage credits to bolster the initiative.
Mythos 5 to Be Reintroduced
On June 12, 2026, Anthropic unexpectedly halted access to both Claude Mythos 5 and Claude Fable 5 for all Glasswing partners. The company announced it was collaborating closely with the U.S. government to investigate the disruption and restore access as swiftly as possible.
The suspension left critical infrastructure protectors without access to the platform’s most advanced threat detection and vulnerability identification functionalities during the review period.
According to Anthropic, as of June 27, 2026, the U.S. government formally informed Anthropic that Claude Mythos 5 can be reintroduced to a defined group of U.S. organizations that operate and protect critical infrastructure.
Anthropic confirmed it is swiftly working to restore access for these verified entities. The company also stated it continues to collaborate with the government regarding a broader expansion of Mythos 5 access and the reinstatement of Claude Fable 5, its most capable general-use model for unrestricted deployment.
This progression carries notable ramifications for the cybersecurity landscape. Defenders working within critical sectors, including energy, healthcare, financial services, and telecommunications, will once again gain access to a model capable of autonomously detecting zero-day vulnerabilities at machine speed.
The phased restoration signals a careful yet accelerating effort by both Anthropic and the U.S. government to leverage advanced AI capabilities for national cyber defense while maintaining strict control over offensive dual-use risks.
“`