Human + agent: Defining identity in the age of the hybrid workforce​ | SailPoint Navigate 2026

BLUF: Securing AI agents requires a unified identity control plane that correlates human and non-human identities, as risks emerge from their intersection rather than from agents in isolation 2:46. SailPoint’s approach addresses this through three pillars: discovery, governance, and protection, ensuring that agent actions are governed by the context of the human owner 14:08.

Key Takeaways:

• The Risk of Fragmented Security: Treating agent security as merely a scale problem leads to failure because breaches often occur when agents exploit human-created vulnerabilities, such as exposed credentials or over-privileged tokens 3:46. The real risk exists in the "space in between" humans and agents, requiring unified visibility 4:45.
• Contextual Risk Scenarios: Disconnected views miss critical risks, such as a low-risk marketing manager using an agent with production write access, or a departing employee leaving active agent tokens behind 7:13. Only a unified view reveals that these combinations create high-severity risks like unauthorized production changes or dormant access 8:40.
• Unified Governance via SailPoint Identity Fabric: The solution maps to three pillars: Discover (finding all agents and their lineage), Govern (managing access rights based on human-agent correlation), and Protect (monitoring for abnormal behavior and enforcing policies) 15:08. Governance ensures agents only have access equivalent to their human owner, preventing privilege escalation 19:28.
• Dynamic Policy Enforcement: Protection involves monitoring prompts, agent plans, and runtime activities to enforce policies dynamically. For example, if a high-risk human uses a high-risk agent, the system can block or flag the action even if the general policy allows it, based on the compounded severity 25:34.

Closing Statement:
Organizations must shift from counting agents to governing the complete identity graph connecting humans and non-human identities. By adopting a unified control plane, enterprises can enable AI productivity while maintaining strict security and compliance postures.

Sources:

• 2:46 Introduction to the problem of treating agent security as a scale issue.
• 4:45 Explanation that breaches live in the space between humans and agents.
• 7:13 Scenario of a marketing manager using an agent with production access.
• 8:40 Risk of departing employees leaving active agent tokens.
• 14:08 Overview of SailPoint Identity Fabric's three pillars.
• 19:28 Governance through correlating human and agent access rights.
• 25:34 Dynamic policy enforcement based on compounded risk severity.