Xage Security Adds Digital Identity Assignment and Policy Enforcement for AI Agents

Xage Security has added AI agent controls to its zero-trust platform, introducing digital identity assignment at onboarding and agent-specific policy enforcement for autonomous software operating across enterprise and industrial networks. The release extends Xage’s identity and access management framework from human users and devices to AI agents.

Under the new system, each AI agent receives a digital identity when it is deployed, with permissions tied to approved business functions and policy controls. The platform is designed to enforce what systems an agent can access and what actions it can take, reducing the risk that an agent manipulated through prompt injection or other adversarial inputs can exceed its authorized scope.

Xage says the platform provides visibility into agent activity, including unmanaged agents, tool usage, data access, and interactions with other systems. Administrators can apply policy controls, detect deviations from expected behavior, block compromised agents, and remove agents that are no longer authorized.

The company is positioning the controls for environments where autonomous agents may interact with operational technology, critical infrastructure, and enterprise applications. In those settings, an agent’s identity, permissions, and audit trail need to be managed separately from the human user or application that originally deployed it.

The release reflects a broader industry push to apply identity governance to non-human actors. Palo Alto Networks launched a platform for human and agentic identity security earlier this month, while SailPoint introduced Agentic Fabric for AI and non-human identity governance. Keycard has also launched identity and access controls specifically for multi-agent AI applications.

Xage, based in Palo Alto, California, has historically focused on zero-trust security for operational technology environments including energy, manufacturing, and defense. The company is extending that model to autonomous systems that need continuous authorization rather than one-time access approval.

Sources: Xage Security

–

By Ji-seo Kim