ServiceNow has announced AI Passport, a new tool that controls what AI agents are authorized to access and do while they are running, according to a company press release.

The Santa Clara, California-based company announced it on October 15 and expects it to be available in the fourth quarter of 2026. ServiceNow executive Jon Sigler said “visibility alone isn’t enough” as AI agents spread across the enterprise, according to the release.

Each AI agent gets its own digitally signed credential, which it checks at runtime to confirm what it is authorized to do, ServiceNow says. The company says the platform can redirect an agent that strays outside that scope, restrict its access or shut it down.

How AI Passport Works

Each credential is issued through ServiceNow’s Super Intelligence Control Tower, the platform the company uses to discover, control and measure AI agents and their spending.

The company says this differs from audit logs, which show what happened after the fact. AI Passport is meant to prevent an agent from acting outside its scope by authorizing what it can do while it runs.

For agents from outside vendors, ServiceNow says AI Passport creates a verifiable permissions record rather than relying on what the agent claims about its own permissions. It also keeps auditable records of who authorized actions, the company says.

NVIDIA is a design and launch partner, building an enforcement implementation with its Open Agent Safety Platform, a set of tools for securing AI agents, according to the release.

Expanded AI Governance Capabilities

ServiceNow is also adding shadow AI detection, which finds unapproved AI agents running on a company’s network, and AI Cost Management, which connects AI spending to business outcomes, according to the release. Both are expected in the fourth quarter of 2026.

Agent Authorization Starts With Knowing What Data It Can Reach

For chief data officers (CDOs), the useful point is that deciding whether an AI agent can act requires knowing what data, tools and systems it is entitled to reach. That makes agent permissions a data governance question as much as a security one.

ServiceNow says each credential is grounded in the customer’s Configuration Management Database (CMDB), which records its IT assets, along with its Governance, Risk and Compliance (GRC) policies. 

The same dependency applies to any organization: AI governance only works when access rules rest on an accurate picture of what data and systems exist.