DNSFilter has released findings from its The Visibility Deficit: The 2026 AI Cybersecurity Report, highlighting a widening gap between how confident executives feel about their organization’s AI and cyber security posture and what their own IT teams are actually experiencing. The study found that this disconnect runs through executive confidence, security budgets, and how well AI tools are governed, and business continuity after a cyber incident is an area of particular dissonance.
Nearly three-quarters of executives (73.9%) believe that their organization would be able to maintain critical operations through a 72-hour cyber attack, compared with just 47.5% of IT management and 50% of cyber security professionals running those systems day to day. Nearly half of IT management (45.8%) say their organization has had unreported security incidents, vs only 17.4% of executives who believe that is the case. While 52% of IT management say incidents have been downplayed due to limited detection capability, 39.1% of executives admit to the same.
The gap is also evident when it comes to AI-related threats. Between 69% and 71% of IT management and network and cloud professionals say attackers are adopting AI faster than their own teams can keep pace, compared with just 43% of executive leaders: more than a quarter of whom (26%) believe their own team is actually ahead. Executives also stay focused on the models they recognise by name, with 65% flagging ChatGPT as a top risk, while practitioners track a far wider, harder-to-govern surface that includes DeepSeek, Kimi K2, and Mistral.
This perception gap is showing up in how spending decisions are made, too. While 74.5% of respondents say that budgets increased over the past year, executives and practitioners are directing that money at different problems. Executives prioritise AI security governance (52.2%) and network security (26.1%), while practitioners want more investment in endpoint detection (22.8%), security training (19.3%), and incident response (18.4%). AI governance is simultaneously flagged as the largest security gap (36.25%) and the top investment priority (40.25%).
Scale introduces its own exposure: the largest organizations run six to ten AI tools at nearly double the overall rate of their smaller peers (47.5% vs 25.2%) and report the highest shadow AI incident rate of any size band, at 32.5% vs 20.1% overall. Shadow AI is already producing measurable damage, with 40% of organizations citing an incident tied to a third-party SaaS or AI tool in the past year, and 20% confirming an unauthorised AI connection directly caused one. More than a third of organizations (36.8%) notify IT of new AI connections only after access has already been granted, and 6.5% send no notification at all.
Additional key findings from the survey include:
- 8.7% of executives see no emerging AI risk at all, more than double the overall rate of other surveyed groups (3.8%).
- 73.9% of executives believe new AI tools require admin approval before access is granted, vs 58-60% of practitioners who say the same.
- Only 49% of organizations have formal, consistently enforced policies governing autonomous AI agents.
Survey methodology
The survey was conducted among 400 IT and cyber security professionals at the manager level and above, at companies with 150 to 5,000 employees, across North America, in June 2026.
