We surveyed 100 CISOs, Deputy CISOs, and security directors to understand what's stopping defenders from moving at machine speed, and what would change their minds.
Inside the Report
Adversaries are already using agents. Defenders mostly aren't. We asked CISOs, Deputy CISOs, and security directors at enterprise and mid-market organizations what's holding them back. The data tells a story of a distrust, with a data visibility problem compounding it
overview
The State of Agentic Security 2026 reports on the results from a survey of 100 cybersecurity leaders examining the gap between security teams who want AI agents in the SOC and those willing to deploy them.
The report identifies trust as the primary barrier to adoption of security agents, with 52% of leaders citing distrust of agent outputs and 64% flagging three or more compounding trust concerns at once. Incomplete data also compounds the distrust of AI agents, with 84% of security teams unable to access all their log data for investigations.
Most security leaders agree agents belong in the SOC, but few have moved beyond evaluation and pilots into production.
Security leaders are flagging multiple trust concerns at once, from hallucination to acting on incomplete data.
Most security teams can't access all their log data, and investigations stall because of it.
the path forward
What security teams can do this quarter to introduce or expand agentic security without migration.
Understand the honest state of agentic security in 2026, and get guidance on what to do now to close the gap between planning to implement agents and actually deploying them.