AI Uncovers Decades of Flaws, Patch Responsibility Remains Unsettled at Infosecurity Europe (infosecurity-magazine.com)

• AI models reduce mean time to exploit from days to under seven hours.
• India mandates 12-hour patches; EU CRA pushes vendor accountability; US relies on market forces.
• Experts call for exploit-intelligence prioritization and assume compromise, not just faster patching.

"At Infosecurity Europe, experts discussed how AI models like Claude Mythos and GPT5.5 autonomously find and fix vulnerabilities. The mean time to exploit dropped to six hours and 40 minutes. India introduced a 12-hour patch deadline for critical flaws; the EU's Cyber Resilience Act shifts producer liability; the US relies on market incentives. Experts argued that patching alone is insufficient, urging exploit-intelligence-driven prioritization and defense-in-depth. The industry is at an inflection point where accountability and operational practice must be renegotiated."