ServiceNow Introduces AI Passport for Enterprise Agents

ServiceNow has introduced AI Passport, a new enterprise governance capability intended to help organisations control the actions of AI agents operating across business systems. The product uses cryptographically signed credentials to establish which actions an agent is authorised to perform. As businesses adopt agents that can retrieve information, update records and trigger workflows, permissions become more complicated than they are for a conventional chatbot. An agent may need temporary access to several applications, but should not automatically inherit unlimited authority simply because it was asked to complete a task. AI Passport is designed to support verification of permitted actions, policy enforcement, access revocation and audit trails. ServiceNow says the approach can work with both its own agents and third-party systems, using open specifications rather than requiring every agent to be built on one platform. The capability is associated with ServiceNow's Super Intelligence Control Tower, which the company positions as a way to discover and monitor AI activity. It also announced AI cost-management features intended to help organisations understand spending across AI tools and infrastructure. ServiceNow is working with NVIDIA on integration with the NVIDIA Open Agent Safety Platform. The collaboration is aimed at strengthening controls closer to the environment where agents execute, rather than relying solely on instructions provided to a model. The announcement distinguishes current offerings from capabilities expected later. According to the company's reported timeline, AI Passport, shadow-AI detection and AI Cost Management are anticipated in the fourth quarter of 2026. Businesses should confirm actual availability and integration requirements before planning deployments. For technology leaders, the key issue is accountability. Organisations need to know which agent performed an action, under whose authority, with access to which data, and how to reverse or investigate a mistake. Credentials and audit trails can help, but they do not replace careful workflow design, testing and human approval for sensitive operations. The development reflects a wider shift in enterprise AI from experimentation toward managed deployment, where security, permissions and operational oversight are essential to adoption.