AI models significantly accelerate exploitation of security vulnerabilities (heise.de)

• Anthropic's Mythos AI found 271 Firefox bugs but only one in Curl.
• A Linux kernel vulnerability was exploited within hours of its patch.
• Monthly patching cycles are obsolete; real-time updates are required.

"AI, particularly Anthropic's Mythos model, can not only find but also actively exploit vulnerabilities. The time between patch release and working exploits has shrunk to hours. Companies must shift from monthly patch cycles to real-time updates. Defenders risk being overwhelmed."