ServiceNow AI Passport Gives AI Agents Signed Credentials

ServiceNow wants every AI agent inside the enterprise to carry a passport. On October 15, the Santa Clara company announced ServiceNow AI Passport, a new authority layer that issues cryptographically signed credentials defining exactly what an AI agent — or the Model Context Protocol (MCP) server it depends on — is authorized to do at runtime. NVIDIA is the design and launch partner, building a production-ready enforcement implementation on its Open Agent Safety Platform.

According to ServiceNow's announcement, the ServiceNow AI Passport is consulted at runtime and verified at every step of a multi-agent workflow. When an agent stays within its authorized scope, it operates freely. When it drifts, the system applies graduated responses: redirecting the agent, restricting its actions, or reconnecting it with proper context — up to kill-switch capabilities that can terminate a rogue agent at machine speed. Behind it all, the company's Super Intelligence Control Tower keeps the full audit record, turning agent execution into what ServiceNow describes as governed, auditable, and accountable work.

A credential the agent checks before every action

The core idea is that enterprise AI agents need more than a capable reasoning model to run production workflows safely. They need structure: which data and tools they can touch, clear rules for what actions are allowed, and an auditable record of who authorized each one, according to the company's announcement.

Each ServiceNow AI Passport credential is issued by the Super Intelligence Control Tower — the product ServiceNow presented as AI Control Tower during a press and analyst briefing on October 13 — and is grounded in live enterprise business context. According to The Source Code, the passport draws on the company's Configuration Management Database (CMDB), governance, risk and compliance policies, and workflow approvals. For third-party agents, the ServiceNow AI Passport creates a verifiable permissions record in place of relying on an agent's own claims about what it is allowed to do.

ServiceNow says the system is built on open specifications, so any governance product, identity provider, or enforcement point can issue and verify passports across multi-agent workflows. The same rules extend to the MCP servers agents use to reach tools and data — an acknowledgment that the agent supply chain now includes infrastructure most enterprises never directly chose.

NVIDIA builds the enforcement layer

NVIDIA joins as the design and launch partner, contributing its Open Agent Safety Platform along with OpenShell runtime software and the Sentry security framework, the Futurum Group reported. In practical terms, NVIDIA is building the production-ready enforcement implementation — the machinery that checks the ServiceNow AI Passport at machine speed and acts when an agent steps outside its scope.

That partnership matters because enforcement is where most agent-governance pitches fall apart. A policy document that says what an agent should do is cheap; a runtime that can redirect, restrict, or terminate the agent mid-workflow is the expensive part. By pairing ServiceNow's authority layer with NVIDIA's runtime, the two companies are betting that enterprises will pay for governance that acts, not just governance that observes.

Jon Sigler, ServiceNow's executive vice president and general manager for the AI platform, framed the announcement around a gap he said visibility tools alone cannot close: as agents proliferate across the enterprise, organizations need confidence that every agent is operating within its authorized scope, and the new operational harness is meant to unify agent authority, governance, and operational oversight so agentic AI can scale with control.

The governance race moves to runtime

The ServiceNow AI Passport announcement arrives alongside a broader expansion of the Control Tower. According to the announcement, the platform gains shadow AI detection to hunt for unapproved agents, more than 400 out-of-box connectors extending discovery across other vendors' platforms, AI Cost Management folded into IT asset management, and continuous enterprise AI observability. AI Passport, shadow AI detection, and AI Cost Management are expected in the fourth quarter of 2026; a separate Operational Resilience solution for financial services is available immediately, reported BigDATAwire.

ServiceNow also introduced the third generation of its employee-centric AI harness, built around AI Agent Studio and Now Intelligence. AI Agent Studio lets employees build agents through natural language with governance controls embedded by default, so IT keeps policy enforcement over employee-created agents, while Now Intelligence delivers contextual, role-specific assistance across the platform, according to the Futurum Group's analysis. The employee harness ships with the Yokohama platform release.

The announcement lands in the middle of a fast-moving governance wave. Microsoft is giving enterprise agents their own credentials through Copilot Studio agent accounts, and regulators are starting to put agentic AI under formal rules, as seen in Hong Kong's banking framework. ServiceNow's contribution pushes the argument from identity to enforcement: it is no longer enough to know which agent is acting — the enterprise needs a credential the agent itself must present, verified at machine speed, before every action, and the ServiceNow AI Passport is ServiceNow's bid to become exactly that.