The Pulse
OpenAI resolves publishing failures in ChatGPT Sites
OpenAI says services are operating normally after publishing failures affected ChatGPT Sites. The company reported elevated errors and work on a mitigation before marking the incident resolved.

Publishing in ChatGPT Sites is back to normal after an incident that caused failures and elevated errors, OpenAI said on its status page. The company marked the incident resolved at 12:55 a.m. on Nov. 13, after reporting the problem on Nov. 12.
Errors and mitigation
At 10:59 p.m. on Nov. 12, OpenAI said it had identified elevated errors affecting the service and was working on a mitigation. The status page listed Sites as the affected component under ChatGPT and classified the incident as degraded performance.
At resolution, the company said all affected services were operating normally again.
What publishing does
Sites lets users create, preview, publish and share interactive websites and lightweight apps from ChatGPT on the web or in its desktop app. The feature is available to ChatGPT Plus and Pro accounts and to Business and Enterprise workspaces.
Workspace settings and role permissions determine who can use Sites. Public publishing is off by default in Enterprise workspaces and must be enabled by an administrator.
New Sites are private by default, although workspace administrators may also have access. Publishing makes the current version available at its URL to the selected audience; deploying changes to an existing Site updates its live version.
Database changes
In its general guidance for Sites, OpenAI says publishing can change a Site’s database before a new version goes live. If publishing fails, some database changes may already have been saved.
Restoring an earlier version restores the code but not the database or uploaded files. That guidance describes how publishing works generally, rather than reporting what happened to any particular Site during this incident.
Sources
Related stories
- OpenAI names Aleph exclusive ChatGPT Ads reseller in 44 markets
- OpenAI Will Pool ChatGPT and Codex Limits in December
- OpenAI Fixed ChatGPT Share-Link Flaw That Exposed Private Chats
- OpenAI Reports More Than 200 Violence Referrals and Tightens Teen Safeguards
- OpenAI Adds Teen Weapon Limits and Pledges a Safety Report