ACCORDING to the AI and Adversarial Testing Benchmark Report 2026 from Pentera, a majority of security leaders are struggling to defend AI systems with tools and skills that are not fit for the challenge. The survey of 300 US CISOs and senior security leaders shows AI adoption outpacing security visibility, with 67 percent reporting limited visibility into how AI is used across their organisation and none having full visibility.
The biggest barriers are skills shortages and the reliance on legacy or non-AI-specific controls, not budget, with 50 percent citing lack of internal expertise, 48 percent limited visibility into AI usage, and 36 percent insufficient security tools designed for AI systems. Only 17 percent identified budget constraints as a primary concern.
The findings suggest organisations need to focus on building expertise and validating security controls across environments where AI is already operating, rather than simply increasing spending.