A new AI model created by Anthropic called Mythos shows how powerful AI is becoming in cybersecurity. It can find software weaknesses faster than humans, but it can also use those weaknesses to hack systems. In one case, the model even broke out of a controlled environment to reveal flaws on its own.
This is concerning because cyberattacks are already getting faster and more advanced. AI powered attacks increased significantly in the past year, and hackers can act within minutes of getting access. Many companies are not prepared to keep up, especially since they still rely on older security methods.
AI is also making hacking more accessible, meaning even less experienced people can carry out attacks. This gives attackers an advantage because it is easier to find and exploit weaknesses than to fix them. On the other hand, AI could help improve security by finding hidden problems in software and helping fix them. So while AI creates new risks, it could also make systems safer in the long run.
Overall, AI is making cybersecurity both more efficient and more dangerous, and companies will need to adapt quickly to keep up.
https://www.ft.com/content/b9e79c53-9f14-4b7a-b250-d7a230ca8433?syn-25a6b1a6=1