ServiceNow Launches AI Passport to Verify AI Agent Permissions in Real Time
October 19, 2026 · 2 min read
ServiceNow has released AI Passport and new advancements in AI governance to ensure that artificial intelligence agents are prepared for production environments. This system combines runtime governance, platform integration, and comprehensive monitoring within the operational AI infrastructure of ServiceNow.
AI Passport introduces an authorization layer that allows companies to control the actions that AI agents and Model Context Protocol servers are authorized to perform during each phase of a multi agent workflow. This system uses cryptographically signed credentials that AI agents consult during runtime, including in environments such as NVIDIA OpenShell, to verify the scope of their authorizations based on business context.
If an AI agent operates within the limits of its passport, it can function freely. If it deviates, the system can apply gradual measures, such as redirecting the agent, restricting its actions, or providing the correct context. These measures include kill switch mechanisms capable of stopping execution at machine speed. The Super Intelligence Control Tower of ServiceNow maintains a full audit log of actions to ensure activity is governed and auditable.
AI Passport is based on open specifications, allowing any governance solution, identity provider, or policy enforcement mechanism to issue and verify credentials for first party and third party AI agents. This approach shifts from reaction to prevention by verifying authorized actions during runtime rather than relying on posterior audit logs. Each agent receives a passport issued by the Super Intelligence Control Tower, which is linked to the configuration management database, governance, risk and compliance policies, and workflow approvals of ServiceNow.
For third party AI agents, this provides a verifiable record of permissions instead of relying on the authorizations the agents claim to have. The system allows for the precise revocation of permissions and the delegation of authorizations with restricted scopes at each stage.
ServiceNow has named NVIDIA as its design and launch partner. The two companies are developing a production ready implementation of these control mechanisms through the NVIDIA Open Agent Safety Platform. This implementation includes the NVIDIA Sentry security framework and the NVIDIA OpenShell open source software for secure execution environments.