What happened
On October 28, 2026, ServiceNow announced an expanded partnership with OpenAI. The stated aim is to connect OpenAI models and agents to ServiceNow workflows through ServiceNow Action Fabric, so teams can move from reasoning to governed action. The release also describes plans for using the companies’ products inside their own operations, including OpenAI’s planned use of ServiceNow and ServiceNow’s plan to make ChatGPT Work and Codex available to its employees.
The concrete, currently available item is a new ServiceNow plugin for ChatGPT, listed in the OpenAI marketplace. It is powered by Action Fabric and combines ChatGPT’s reasoning with ServiceNow’s enterprise context, business rules, and permissions. The release gives an IT example: an agent can find similar incidents, assess change risk, and update applicable records without leaving ChatGPT. ServiceNow remains the system where those actions are executed and governed. The announcement does not give a price, a customer eligibility list, or a detailed setup procedure. It says the plugin is available today and that enhanced capabilities are expected by the end of 2026.
The remaining announcements are future work with separate schedules. ServiceNow and OpenAI are collaborating on security and risk capabilities intended to help teams identify, triage, validate, and mitigate attack paths across code, applications, workloads, networks, and identities. The companies describe combining OpenAI cybersecurity capabilities with ServiceNow asset, access, and enterprise context. Early access is expected in January 2027, with general availability planned for the second quarter of 2027. The release notes that codebase scans can surface 10,000 to 100,000 existing and synthetic vulnerabilities or more, and frames the challenge as prioritizing imminent risk—not as a measured result of the announced product.
They also plan to integrate Codex with ServiceNow AI Workflow Factory, where OpenAI models are also expected to power Build Agent and Autonomous Engineer. Codex in AI Workflow Factory is expected in 2027. Separately, ServiceNow plans a voice experience for its Super Intelligence Control Tower: it would collect an AI-use-case description, pre-fill risk-assessment fields, ask follow-up questions when information is missing or unclear, and assign a risk rating based on the stated purpose. That voice capability is expected in the first quarter of 2027. OpenAI is also joining ServiceNow’s Control Tower Alliance as a founding member. The release supplies no measured productivity, security, or cost outcomes for these plans.
What the archive adds
This is a new development on a known partnership story. OpenAI and ServiceNow expand partnership across enterprise workflows covered the earlier expansion; this announcement adds the available ChatGPT plugin and schedules for security, voice, and Codex work. An earlier Telegram post also described the partnership, but this is not a link to that post.
Trend signal
The announcement supports agent-ready software exposes capabilities: external agents can invoke bounded enterprise actions while the host platform retains context, permissions, and execution governance. The release describes the intended boundary, not independent evidence that it works across customers or that it is safe in every deployment.
Why now
The release presents the expanded partnership as a move from model reasoning toward governed action in enterprise workflows. The timing is also concrete: the plugin is available now, while other capabilities have announced 2026–2027 schedules. That split makes this an implementation announcement as well as a roadmap update.
Why it matters
For teams integrating agents with systems of record, the key architectural choice is where authority lives. In ServiceNow’s description, ChatGPT can interpret a request and select a workflow, but the platform supplies business context and permissions and remains the place where records are changed. That separation is a useful design constraint: an agent interface need not become the system of record or receive unrestricted access to its data.
The plugin could give employees a conversational route into familiar IT processes, but the announcement does not establish how broadly it can be configured, what actions require approval, or how administrators audit each operation. Teams evaluating similar integrations should map actions to roles, permission scopes, approval gates, and logs before treating a chat interface as operational automation. The release is evidence of a shipping integration, not proof of reduced incident-resolution time or improved security outcomes. Earlier coverage: OpenAI and ServiceNow expand partnership across enterprise workflows.
What happens next
- ServiceNow will expand plugin capabilities beyond the initial marketplace integration. (3 months, 70%). Watch for: a release describing the enhanced capabilities expected by the end of 2026.
- The planned voice risk-assessment experience will enter preview or availability. (6 months, 60%). Watch for: ServiceNow’s announcement against its first-quarter 2027 schedule.
Limits and uncertainty
The source is ServiceNow’s partnership announcement, not an independent product evaluation. It does not specify pricing, customer or regional availability constraints, configuration requirements, supported ServiceNow workflow types, model versions, or the exact permissions and approval policies available to administrators. The IT incident example is illustrative; no deployment results or performance measurements are reported. The cybersecurity, Codex, and voice features are future plans with target dates, not capabilities to assume are usable today. Nor does the announcement establish that OpenAI’s internal adoption plans have been completed.
Skill to build now
Before connecting any chat or agent interface to a workflow platform, make a small action inventory: list the proposed actions, the records each action can read or change, the role allowed to invoke it, and which changes need approval. For each action, define an audit event and a safe test case. That exercise is directly relevant to this plugin’s model—chat-based reasoning paired with platform-enforced context and permissions—and can expose whether the integration boundary is concrete enough to review.
Reader takeaway
The most actionable part of the announcement is already available: a ServiceNow plugin in the OpenAI marketplace for governed workflow actions from ChatGPT. Treat it as a way to put a conversational front end over controlled enterprise execution, not as evidence that the broader security, voice, or Codex roadmap has shipped or produced measurable results.