AI-assisted security tools are finding more bugs, but the threat level has not changed

VulnCheck's H1 2026 analysis found 1,061 AI-assisted vulnerabilities with a 1.3% exploitation rate—comparable to the overall rate—while average time-to-exploitation dropped from 120 days in 2025 to 80 days; content management systems, network edge devices, operating systems, server software and emerging AI products were the most commonly exploited categories. The report concludes AI-discovered vulnerabilities are not yet more likely to be exploited than traditionally discovered flaws but warns the risk could grow as AI vulnerability-hunting models are more widely deployed.