
Intruder survey finds 55% of organizations use AI pentesting, with data privacy and production safety as top barriers
Intruder's survey of 500 security and engineering leaders shows AI pentesting is spreading fast, but adoption is being gated by data privacy, production safety, and auditability rather than bug-finding quality alone. The findings matter because teams shipping code daily are already accepting gaps between tests and spending hours validating security tool output.
Intruder said 55% of surveyed organizations use AI pentesting across their environment, with 36% having tried it without scaling and fewer than 1% having no plans to try it. The survey covered 500 security and engineering leaders in the United States and Britain, a group responsible for deciding whether automated offensive security tools can run against live web applications and APIs. The report says time-to-exploit has fallen to a single day, while adoption was highest among the largest organizations, where 71% reported using AI pentesting environment-wide.
The data show why teams are looking beyond scheduled tests: 68% deploy code to production at least daily and 37% deploy multiple times a day, while 47% leave more than a month between full pentests on their most frequently tested web applications and APIs. Only 4% believe current testing frequency gives sufficient coverage, and 42% accept shipping code without additional security checks between pentests. The survey also found 88% of leaders running scheduled pentests worry about unmapped risk introduced between tests.
Context matters because AI pentesting moves from lab-style assessment into live systems. The survey placed data privacy first at 40%, production safety second at 36%, and auditor or regulator acceptance third at 28%, while only about a quarter named testing quality as a top concern. That suggests the limiting factor is not whether AI can find bugs, but whether security teams can control where it looks, what data it sees, and how it proves findings. Nearly half said missing visibility or non-reproducible exploit proof would damage trust.
The operational stakes fall on organizations that ship fast but cannot afford an accidental outage or data exposure from a testing agent. AI pentesting is presented in the survey as a way to run in-depth tests whenever teams need them, yet 56% of teams already spend six or more hours a week validating security findings, and one in five spend more than eleven hours. The survey also found 44% call validating findings the biggest remediation bottleneck, and 46% said missing visibility or non-reproducible exploit proof would damage trust.
It affects security and engineering teams that deploy code daily: AI pentesting is being adopted to close testing gaps, but the survey says data privacy, production safety, and auditability are the leading barriers.
Sources (1)
- [1]Original source(https://thehackernews.com/2026/11/quality-isnt-main-worry-about-ai.html)