AI Security & Exposure Survey 2026: What CISOs Say They’re Missing - LinkedIn

AI adoption is widespread. AI security governance isn’t. AI is now in use across 100% of enterprises - but security governance hasn’t kept pace. Pentera’s AI Security & Exposure Survey 2026 shows that AI adoption has outpaced security teams’ ability to maintain visibility, establish ownership, or test systems under real-world adversarial conditions. Where AI Security Falls Short The Report highlights several consistent challenges across organizations: Skills and expertise gaps: 50% cite a lack of internal expertise as a top barrier to securing AI systems. Reliance on legacy controls: 75% of CISOs report extending security controls originally designed for other attack surfaces to cover AI-driven workflows and infrastructure. Limited AI visibility: Nearly 70% of CISOs report limited insight into how AI is being used across their environments, making it difficult to understand exposure and risk. AI tooling and testing limitations: 90% of CISOs report not having security tools designed specifically to protect AI systems, while a ⅓  say their AI testing approaches remain unclear. NEW: Download The AI Security Report 2026 Recommended by LinkedIn Action at Scale Robotic Assistance Devices 8 months ago Part 1: The Many Ways LLMs Leak Data—and How to Solve… Cota Capital 11 months ago 🔐 Industry Experts Share DSPM, Automation, and AI… CodeGuardian.ai 1 year ago Why This Matters The message is clear: AI adoption is moving faster than organizations’ ability to secure it. When AI environments lack visibility, purpose-built controls, and adversarial testing, they become exposed attack surfaces with limited validation. Security teams know AI is present - but without adversarial testing, they can’t see how attackers might exploit AI-based connections across data, identity, and infrastructure. Join The Live Session: Closing The AI Security Gap Deep dive into what 300 CISOs reveal about AI security & Exposure - and benchmark your organization against peer enterprises. Join us for a live session on March 11, 11:00 am ET, where Ryan Dorey, Director of Technical Advisors at Pentera, and Chris Cochran, Field CISO & VP of AI Security at SANS, will: 📌  Break down key benchmark insights 📌 Share examples of exploitable risk caused by limited AI visibility and reliance on legacy controls 📌 Show how adversarial testing helps security teams validate AI risk Register for the webinar HERE