Probably Not Exploitable
ProjectDiscovery surveyed 214 security practitioners and found that 44% close or deprioritize findings without full verification due to capacity constraints, with 72% worried about real vulnerabilities sitting unnoticed in backlogs. Security leaders (CISOs) admit to this practice most openly (54%), and backlogs grew for 74% of teams over the past year. The survey reveals AI is paradoxically both exacerbating the problem (by generating more findings) and helping solve it (enabling faster triage), with teams using AI triage reporting better backlog sentiment. The core challenge is that vulnerability queues grow faster than teams can process them, making old assumptions like “probably not exploitable” riskier now that attackers can validate exploits at scale using AI—though better-designed AI triage systems with contextual memory could help teams focus on high-signal issues rather than adding noise.