Intense concerns in cybersecurity cause reports on the potential of the artificial intelligence model Mythos of Anthropic which allegedly managed to penetrate highly protected information systems within a few hours.
According to claims released in technological and security circles, the artificial intelligence system reportedly violated almost all of the NSA's classified information systems (US National Security Agency) in a few hours and not weeks.
These reports are attributed to the head of the American cybersecurity agency, who allegedly argued to Economist that the model «He invaded almost all our classified systems, not in weeks, but in hours».
BREAKING: The NSA confirms Mythos "broke into above all of our classified systems, not in weeks, but in hours"
— Polymarket (@Polymarket) June 21, 2026
These claims have raised debate as to whether new models artificial intelligence can be developed into tools capable of automatically detecting and exploiting weaknesses even in high protection networks.
Concern is reinforced by the general debate on the safety of artificial intelligence and the risk of advanced systems being used for large-scale cyberattacks.
What Anthropic Mythos is
Mythos is a new, very powerful artificial intelligence model of Anthropic, the company created by Claude AI. Unlike a simple chatbot, it is designed for more complex tasks, such as code analysis and troubleshooting in computer systems.
The interest around Mythos It's mostly about cybersecurity. This artificial intelligence can help experts find weaknesses in networks and programs before they exploit hackers.
However, the same possibility raises concerns, as such a powerful tool could theoretically be used for malicious attacks. That's why Anthropic tests it with limited access and not as a simple public tool.
In simple terms, Mythos is one «digital security expert» with very large possibilities, which can help protect systems, but at the same time creates questions about how powerful artificial intelligence should be used.
