Key Findings
38% of security leaders are very confident they can prove what an AI agent did through audit records.
79% of security leaders expect their organization will need to claw back or significantly restrict AI agent usage within the next 18 months.
37% of security leaders monitor AI agent activity very closely.
80% of security leaders are concerned that AI agents may hold access that was never explicitly granted.
85% of security leaders say project-level isolation is essential or very essential for AI adoption.
51% of security leaders rank project isolation as a desired control before expanding agent fleets.
48% of security leaders rank session logging as a desired control before expanding agent fleets.
87% of security leaders agree that authorization is the missing layer in agentic AI security.
43% of security leaders are very confident they can demonstrate what AI agents were explicitly authorized to do.
36% of security leaders strongly agree that company leadership has directed the deployment of AI agents or agentic capabilities.
75% of security leaders are concerned that an AI agent could perform the correct task in the wrong environment or location.
84% of security leaders believe agents can cross project boundaries more easily than employees.
56% of security leaders rank explicit authorization as the top control they want before expanding agent fleets.
58% of organizations operate more than 50 AI agents today.
66% of organizations expect to operate more than 50 AI agents within 12 months.
38% of organizations expect to operate more than 100 AI agents within 12 months.