AI Vulnerability Discovery Outpaces Patching Efforts

• •AI has identified over 10,000 critical vulnerabilities, creating a backlog for patching.
• •Mean time to exploit has dropped to zero, allowing attackers to exploit vulnerabilities before patches are available.
• •AI tools can generate exploits in minutes, while traditional patching can take weeks.

Artificial intelligence has significantly improved at identifying software vulnerabilities, with Anthropic's Claude Mythos Preview uncovering over 10,000 high- or critical-severity flaws. This rapid discovery poses challenges for organizations as traditional patch management processes struggle to keep pace, with mean time to exploit dropping to zero in 2024. AI can generate working exploits for disclosed vulnerabilities in just 10 to 15 minutes, while defenders may take weeks to deploy fixes. The emergence of AI-powered vulnerability clearinghouses is reshaping the landscape of vulnerability management, but it also creates risks as published patches can inform attackers about changes in software. Organizations are urged to align their security programs with actual exploitability rather than just patch deployment speed.