Identity Security 2026: 4 Predictions & Recommendations

Hackers Target Agents and High-profile Security Incidents Crop Up as Agentic AI Moves from Enterprise Experiment to Production

Agentic AI is transitioning from experimental deployments within SaaS environments to impacting core business processes, increasing cyber risk. Omdia research in 2025 indicated widespread AI agent deployment, but primarily within secure “walled gardens.”

Why This Matters

While agentic AI promises productivity gains, its integration with core applications exposes significant vulnerabilities. Existing security models often struggle to adapt to the dynamic nature of AI agents, potentially leading to data breaches and fraud. A single data breach can cost organizations millions in remediation and reputational damage.

Key Insights

• 11 person-hours: Average time to investigate and remediate a critical identity-related security alert (Omdia, 2025).
• Sagas over centralized transactions: Agentic AI often requires distributed systems patterns like Sagas to manage complex workflows, increasing complexity for security teams.
• ISPM/ITDR: Identity Security Posture Management and Identity Threat Detection and Response are emerging categories to address AI-related security gaps.

Practical Applications

• Financial Institutions: Utilizing AI agents for fraud detection while needing robust identity governance to prevent agent manipulation.
• Pitfall: Over-permissioning AI agents grants excessive access, creating a high-impact single point of failure.

References:

• https://www.darkreading.com/identity-access-management-security/identity-security-2026-predictions-and-recommendations
• https://omdia.tech.informa.com/ (implied from context)