Identity security provider Delinea has introduced runtime authorization capabilities designed to control what AI agents can do after they connect to sensitive enterprise systems.
The Delinea Platform evaluates individual tool calls, commands, and database queries as they occur, allowing organizations to approve, block, or escalate agent actions instead of relying solely on authentication at the start of a session.
AI agents create new privileged-access risks
The release comes as autonomous AI agents increasingly handle tasks across critical corporate IT infrastructure, including production databases, cloud consoles, SSH hosts, Kubernetes clusters, and Model Context Protocol (MCP) servers.
Traditional privileged access management tools were engineered around human workflows, in which verifying a user’s credentials at login was usually sufficient to establish trust. However, autonomous agents can execute dozens of tool calls in seconds, introducing unique operational risks even when properly authenticated.
For MSPs and MSSPs overseeing AI deployments across multiple customer environments, the capabilities could provide a way to apply consistent access policies, human approval requirements, and audit controls without giving agents standing credentials.
As Channel Insider has reported, every API, data source, and tool connected to an autonomous agent can introduce another machine identity that partners must govern.
Delinea applies policies inside active sessions
Delinea’s action-level controls align with the resulting need for short-lived privileges and infrastructure-level enforcement, while potentially supporting recurring identity security services as the company expands its partner program for managed service providers.
“The security industry spent years solving credential theft, but AI agents have introduced a new problem: authorized access doing unauthorized things,” said Art Gilliland, CEO at Delinea.
“You can have perfect credential hygiene and still watch an agent tear through your production environment in milliseconds. Recording what happened or revoking access after the fact doesn’t stop that – enforcing policy on the action as it runs does. The perimeter has moved to inside the session, and no one has figured out how to address that until now.”
To address this shift, the Delinea Platform determines whether a connection request originates from a human or an AI agent before granting entry. This distinction allows security teams to apply agent-tailored policies immediately upon connection, ensuring safety guardrails are active before the agent’s first execution.
Just-in-time credentials limit standing access
Under the new capabilities, AI agents operate without standing credentials. Delinea delivers credentials on a just-in-time basis at the precise moment of access, scoping permission strictly to the required task rather than inheriting broad user privileges.
Once the assigned task completes, the system automatically revokes the credential to prevent persistent vulnerability.
Within an active session, the platform reviews each tool call, database query, or command. Security policies can automatically allow an action, block suspicious behavior, or escalate the step for human approval before execution occurs.
Additionally, the system maintains an audit trail attributing every agent-driven action to a named identity across all connection protocols. This centralized enforcement point provides security teams with visibility across diverse deployment environments without requiring separate governance infrastructure.