ServiceNow Plugin Arrives in ChatGPT as OpenAI Partnership Widens

ServiceNow customers can now connect ChatGPT to their own ServiceNow instance through an MCP plugin that went live in OpenAI’s marketplace on 28 October. It is the first working product of an expanded partnership between the two companies, and for now the only one: a joint security product, a voice feature and Codex inside ServiceNow’s app builder are all slated for 2027.

OpenAI announced the expansion in a post saying ServiceNow “will bring OpenAI models and agents into workflows from development to security, and expand ChatGPT Work and Codex to employees.” It builds on a multi-year agreement from January, when ServiceNow said OpenAI models would be “a preferred intelligence capability” for its enterprise customers and that it would build speech-to-speech technology on them. No financial terms have been disclosed.

The timing leaves ServiceNow courting both big assistant makers in the same week. A day earlier it plugged its industry workflows into Google’s Gemini Enterprise.

Only the ChatGPT plugin is available today

The live piece is the ServiceNow MCP plugin. MCP, the Model Context Protocol, is a common way for AI assistants to call outside tools. OpenAI says the integration brings “business context and workflow actions directly into ChatGPT”, so an IT worker could find similar incidents, assess the risk of a change and update a ServiceNow record without leaving the chat. OpenAI expects “enhanced capabilities” by the end of 2026.

Control stays with the customer’s existing set-up. Customers choose which capabilities to switch on, and ServiceNow’s Action Fabric “applies existing business rules and permissions, so the actions people can take reflect their organization’s configuration and access”, according to OpenAI. ServiceNow is also previewing interactive experiences inside ChatGPT. In OpenAI’s example, an on-call engineer reviews linked monitoring alerts, recent changes and affected services, then assigns the right team. No release date is given for that preview.

Security, voice and Codex follow in 2027

Everything else is a roadmap, and the dates are OpenAI’s estimates.

The security product would combine OpenAI’s cybersecurity capabilities with ServiceNow’s asset, identity and business data to help teams investigate vulnerabilities, prioritise risks and co-ordinate fixes. The voice feature would let people describe an AI use case aloud in the Control Tower, ServiceNow’s console for overseeing AI, with follow-up questions to fill in missing details. The Codex integration would help developers build applications in AI Workflow Factory using ServiceNow’s business context, and OpenAI models will also power ServiceNow’s Build Agent and Autonomous Engineer tools.

The shape will look familiar. Oracle’s expanded pact with OpenAI, announced a day earlier, also paired ChatGPT plugins with a security service built on OpenAI’s cyber models.

Every ServiceNow employee is due to get ChatGPT Work

Inside its own walls, ServiceNow “plans to give all employees access to ChatGPT Work and Codex”, widening use beyond its engineering organisation. OpenAI says engineers can use Codex to explore codebases, make changes and run tests, and that ServiceNow also plans to use OpenAI tools in its internal security work. The traffic runs both ways: OpenAI says it uses ServiceNow workflows in its own operations, supported by the Control Tower.

Neither company says how many people the rollout covers or when it will be complete. Uber’s move to ChatGPT Work for more than 30,000 staff last week followed a 1,000-person pilot.

Permissions, not prompts, decide what the agent can do

The most consequential line in the announcement is the least flashy one: actions taken from ChatGPT run under ServiceNow’s own permission rules. When an assistant can update incident records or push a change forward, that is the difference between a useful tool and an unvetted admin account. It is also the architecture ServiceNow has pushed this month, from AI Passport credentials that limit what agents and MCP servers may do at runtime to its OpenShell-enforced Arc. If the rules live inside ServiceNow, the assistant on top becomes interchangeable, which helps explain how ServiceNow can sign with Google and OpenAI on consecutive days.

For compliance teams, that design helps without settling the hard questions. Existing access rules are only as sound as their last review, and an agent working at chat speed will find an over-broad role faster than any person would. Audit trails, change approval and segregation of duties will need testing once the plugin is switched on, rather than being taken on trust from a launch post.

The other caution is how much of this is still promise. One product ships today. The security, voice and Codex pieces sit between January and an unspecified point in 2027, and voice was already part of the January agreement. Buyers should weigh what is in the marketplace, not what is previewed.

The earnings call is the first test

ServiceNow confirmed the expanded partnership on its own side on 28 October, in a post from its newsroom account on X, the same day it held its third-quarter earnings call. The next fixed marker is January 2027, when the security product is due to open for early access.

Sources

• OpenAI, “OpenAI and ServiceNow expand partnership across enterprise workflows”, 28 October 2026
• ServiceNow, “ServiceNow and OpenAI collaborate to deepen and accelerate enterprise AI outcomes”, 20 January 2026
• ServiceNow News on X, 28 October 2026
• ServiceNow News on X, 27 October 2026

Related stories

• ServiceNow Plugs Its Industry Workflows Into Gemini Enterprise
• NVIDIA Launches OpenShell Blueprint for Safe AI Agents as ServiceNow Ships Arc
• ServiceNow Announces AI Passport to Limit What AI Agents and MCP Servers May Do at Runtime
• Oracle and OpenAI Expand Pact With 4 Products, From ChatGPT to Codex
• Uber Rolls Out ChatGPT Work to More Than 30,000 Employees