InfoSecBulletin Cybersecurity for mankind
Anthropic said that Claude Mythos 5, its strongest AI security model, will be sent back to some U.S. orgs that manage and protect important systems. This comes after a government review that started on June 12, 2026.
The model showed a unique skill to find software problems on its own, revealing thousands of serious flaws in all major operating systems and web browsers.
Hackers Actively Exploiting Palo Alto VPN Flaw Via CVE-2026-0257
DeepSeek AI banned by U.S. Government; federal agencies must remove all related tools
CISA, NSA and NIST Publish New 10 AI Agent Control Guidelines
Claude AI Cracks 25-Year NSA Cryptography Number Mystery in Just 30 Minutes
Bypassing firewalls, AI Agents Accessed Indian ‘ Data Without Permission: CERT-In
MITRE ATT&CK v20 Unveils 9 New Attack Techniques On AI Exploits, MFA Bypass, and CAPTCHA Evasion
Hackers Exploit Fortinet Flaws to Deploy Stealthy PEERSTITCH Malware on FortiGate Firewalls
Hacker stole Meta GitHub repos; Stole Source Code
Urgent patch! Zoom,NVIDIA and IBM Patch 41 Security flaws
Microsoft patches record braking 1411 flaws; 2 zero days
A 27-year-old weakness in OpenBSD and a 16-year-old problem in FFmpeg were found, as well as links in Linux kernel issues that could lead to complete control.
The model is said to have a 72% success rate in making working exploits and linking vulnerabilities on the first try — while the old Opus model had a 0% success rate.
Anthropic started Project Glasswing, a program that only about 200 approved groups could join. This included U.S. government offices and big tech companies like Apple, Amazon, and Microsoft. They promised to spend up to $100 million in credits for using Claude to help this project.
Mythos 5 to Be Redeployed
On June 12, 2026, Anthropic suddenly stopped access to Claude Mythos 5 and Claude Fable 5 for all Glasswing partners. The company said it was partnering with the U.S. government to look into the issue and fix access as fast as it could.
The suspension stopped important infrastructure defenders from using the platform’s best tools for finding threats and identifying problems during the review period.
According to Anthropic, on June 27, 2026, the U.S. government told Anthropic that Claude Mythos 5 can be used again by certain U.S. organizations that protect important services.
Anthropic confirmed it is moving quickly to restore access for these validated entities. The company said it is still working with the government to expand access to Mythos 5 and to bring back Claude Fable 5, its best model for general use.
This change is very important for people in cybersecurity. Defenders in key areas like energy, healthcare, finance, and telecommunications will have a new tool that can find hidden security flaws quickly on its own.
The step-by-step restoration shows a careful but quick effort by both Anthropic and the U.S. government to use advanced AI for national cyber defense while keeping offensive risks under strict control.
