The Hacker News
Intruder report finds privacy and safety concerns in AI pentesting
Intruder’s state-of-ai-pentesting report, based on a survey of 500 security and engineering leaders across the US and UK, finds that the pace of software delivery driven by AI is not just accelerating development but also the window for vulnerabilities to reach production. The article notes that only 4% of security leaders believe their current testing frequency provides sufficient coverage, with 68% of organisations deploying to production at least daily and 37% multiple times a day. Between pentests, however, nearly half (47%) run full tests less than once a month for their web apps and APIs, creating a gap that increases risk.
The findings highlight concerns centred on how AI testing operates in sensitive environments rather than on AI testing quality. Data privacy is identified by 40% as a top barrier, safety on production systems by 36%, and regulatory acceptance of AI-pentest reports by 28%. When asked what would most damage trust in AI pentesting, 46% point to insufficient visibility into what was tested or lack of reproducible proof of exploitability.
Adoption is rapid: 55% of organisations already use AI pentesting across their environment, 36% have tried it but not scaled, and fewer than 1% have no plans to try it. The largest organisations (5,000–10,000 employees) report the highest uptake at 71%. The piece attributes these figures to Intruder’s survey and emphasises the shift towards continuous AI pentesting as standard practice to keep pace with development and attack speed.