Claude Mythos discovers vulnerabilities in classified US government systems
Officials have revealed that Anthropic's AI 'Claude Mythos,' which is known for
Anthropic test found vulnerabilities in classified US systems in hours | AP News
https://apnews.com/article/anthropic-mythos-ai-classified-systems-vulnerabilities-testing-3e8762c0527c4d8ed657cbe48c84a718
Anthropic's Mythos found flaws in classified US systems during a government test
https://thenextweb.com/news/anthropic-mythos-classified-systems-vulnerabilities
Mythos discovers 'Squidbleed,' a memory leak that's gone undetected since Clinton era
https://www.theregister.com/security/2026/06/23/mythos-discovers-squidbleed-a-memory-leak-thats-gone-undetected-since-clinton-era/5260367
Even in its preview version, 'Claude Mythos' was noted for its high cyberattack capabilities, and verification by a UK government agency revealed that it could autonomously execute a complete network takeover attack.
Tests by a UK government agency revealed that 'Claude Mythos Preview' can autonomously execute a complete network takeover attack - GIGAZINE
According to an anonymous government official who spoke to the Associated Press, Claude Mythos also discovered vulnerabilities in classified U.S. government systems. However, these were discovered as part of 'Project Glasswing,' an initiative launched by Anthropic to identify vulnerabilities before they can be exploited by attackers, and the system was not compromised by an external intrusion.
The U.S. government
Anthropic has not disclosed what vulnerabilities were found, and the relevant organizations have not commented, so the specific details remain unknown. The person in question reportedly said, 'We were able to discover a specific vulnerability within a few hours, but that does not mean that Claude Mythos could immediately exploit it.'
Related Posts:
