Ping Identity Adds Runtime Access Controls for Claude Personal Agents

Ping Identity has launched Enterprise Personal Agent Access, a PingOne Privilege capability designed to discover desktop and coding agents, connect their activity to a verified user and device, and enforce access policies while the agents operate.

The first named use cases include Anthropic’s Claude desktop assistants and Claude Code. Ping says the capability is available now and is being piloted with global enterprises seeking to manage personal agents that employees install or use outside centrally approved workflows.

Discovery is intended to identify both approved tools and shadow agents. Once an agent session is associated with a person and device, administrators can allow, deny or log individual actions, require human approval for sensitive tasks and revoke access in real time.

The product applies those controls when agents connect to resources including Model Context Protocol servers, software repositories, APIs, Kubernetes clusters, databases and cloud infrastructure. Policies can account for the user, device, agent, target resource and requested action rather than treating the agent as an independent service account.

For software development, Ping says agents can reach approved resources without being given long-lived credentials. The system can also distinguish a developer’s work from an agent’s actions when attributing code commits, preserving a more detailed audit trail.

The approach is intended to address a gap created when AI assistants act with the permissions of an employee but are not visible to an organization’s existing identity inventory. Runtime enforcement gives security teams a point at which to evaluate each request before an agent reaches a protected system.

Enterprise Personal Agent Access extends Ping’s broader Identity for AI platform, which introduced controls for registering agents, governing their permissions and authorizing machine-driven transactions. The new capability focuses on agents tied directly to an individual employee’s desktop and development environment.

Identity vendors have introduced a wave of controls for AI agents as enterprises move assistants from text generation into tasks that can change code, retrieve sensitive data or operate cloud services.

PingOne Privilege uses short-lived, policy-controlled access in place of credentials stored by the agent. That model is intended to preserve the employee’s accountability while allowing administrators to constrain what the software can do on the employee’s behalf.

Sources: Ping Identity, Ping Identity for AI

–

By the ID Tech Editorial Team