Sysdig Report Warns Vulnerability Growth Is Outpacing Security Teams

Summary

Sysdig’s 2026 report says exploitable in-use vulnerabilities among its users fell 75% year over year, but vulnerability growth is outpacing teams. The article urges stronger runtime defenses and agentic AI remediation under human-set guardrails.

Key points

Sysdig’s 2026 Cloud-Native Security and Usage Report says exploitable in-use vulnerabilities among its users fell 75% year over year.
The article says in-use vulnerabilities without a known exploit have plateaued at 5%, while the overall volume of vulnerabilities continues to grow.
Citing VulnCheck, it reports that vulnerability weaponization timelines fell from nearly a year in 2018 to eight days in 2023; it also cites exploitation of React2Shell within hours of disclosure and CVE-2026-39987 in under 10 hours.
The article warns that AI could further shorten weaponization timelines, making the absence of a known exploit an unreliable safety signal.
The report says more than 70% of organizations use behavior-based detections across 91% of environments, and 140% more organizations auto-kill processes when detections trigger.
Sysdig advocates agentic AI remediation governed by human-defined policies, scoped permissions, rollback plans, audit logging, explainability, and risk thresholds.

Article Details

Publisher: Sysdig
Report Period: 2026
Scope: Vulnerability management and automated response adoption, including findings among Sysdig users and organizations.
Key Statistics: Sysdig reports a 75% year-over-year reduction in exploitable in-use vulnerabilities among its users.
In-use vulnerabilities without a known exploit have plateaued at 5% since the previous year.
CVE-2026-39987 reportedly took less than 10 hours to weaponize, with no proof of concept used as a reference.
More than 70% of organizations use behavior-based detections across 91% of environments, according to the report.
The report says 140% more organizations auto-kill processes when a detection is triggered.
Recommendations: Use runtime security and automated response alongside vulnerability prioritization.
Govern agentic AI remediation with change limits, scoped permissions, deterministic rollback plans, audit logs, explainability, and risk thresholds informed by environmental context.