AI finds plenty of security flaws, but almost none of them get exploited
VulnCheck tracked how often security flaws discovered by AI are actually exploited. Of 1,061 AI-discovered vulnerabilities in the first half of 2026, only 14 saw confirmed attacks—a rate of 1.3 percent, which matches the overall exploitation rate. However, the median time to exploitation has dropped from 120 days to 80 days.
Why it matters: This suggests that while AI-discovered vulnerabilities are not more likely to be exploited, they are being weaponized more quickly, potentially impacting how security teams prioritize their response.
Full story at: The Decoder ↗