AI is now exposing a problem in cybersecurity that companies are struggling to keep up with.
Anthropic’s new AI model, Mythos, is reportedly able to find software security flaws much faster than companies can fix them.
The model is designed to scan systems like operating systems, browsers, and core software tools to detect hidden vulnerabilities that could be used for attacks.
In testing, it has already discovered a large number of serious security issues, showing how many weaknesses still exist in widely used systems.
The main concern is speed.
AI is now finding problems faster than human teams can patch them, creating a growing gap between detection and response.
Security experts say this could make it harder for companies to stay ahead of potential attacks, especially as AI tools become more widely available.
To manage this, Anthropic is limiting access to the system so that companies can fix issues before attackers can take advantage of them.
But the bigger issue remains.
If AI keeps improving at this rate, cybersecurity teams may have to deal with a constant flow of new vulnerabilities instead of occasional fixes.
This changes the way security work is done.
It is no longer just about protecting systems from hackers, but also about keeping up with machines that can find weaknesses at scale.
