AI finds thousands of old software vulnerabilities

Anthropic's AI model, Mythos, has autonomously discovered thousands of previously unknown software vulnerabilities, including some that have persisted for decades. This AI's capability far surpasses traditional security tools and human review, demonstrating a significant leap in exploit detection. The findings span across major operating systems and browsers, highlighting critical flaws in long-standing security assumptions and detection methods. This breakthrough necessitates a fundamental shift in cybersecurity strategies, moving beyond current detection playbooks to address the speed and scale at which AI can now uncover and potentially exploit system weaknesses.