Report by Exabeam

The Agentic Insider: From Monitoring to Understanding

9 FINDINGSPublished Sep 16, 2026
View Original Report →

Key Findings

56% of security leaders use behavioral monitoring and baselining to monitor AI agents.

Behavioral MonitoringAnomaly DetectionAgentic AI

27% of security leaders identify limited behavioral context and correlation as the biggest limitation of their current AI agent monitoring approach.

Behavioral MonitoringMonitoring LimitationsAgentic AI

93% of security and finance leaders say the two functions are aligned on cybersecurity risk tolerance.

Risk Tolerance

48% of security leaders rank AI agents operating with excessive, compromised, or unintended access as the greatest threat to their organization, while 28% rank external threat actors, 12% rank compromised insiders, and 12% rank malicious insiders.

Agentic AIInsider RiskAccess Control

29% of security leaders continue to rely on manual review to monitor AI agents.

Manual ReviewSecurity MonitoringAgentic AI

60% of security leaders use dedicated AI security or governance tooling to monitor AI agents.

AI SecuritySecurity MonitoringAgentic AIAI Governance

56% of security leaders extend existing SIEM, detection, or monitoring platforms to monitor AI agents.

Security MonitoringSIEMAgentic AIMonitoring

81% of security leaders say their CFO understands the cybersecurity risks they are working to mitigate.

Cybersecurity Risks

55% of security leaders delay or scale back a security initiative because they cannot frame the risk in financial terms their CFO will accept.

Security InvestmentFinance AlignmentCybersecurity Risks