OpenAI Outage History

OpenAI experienced elevated failures affecting some users attempting to join ChatGPT Enterprise and ChatGPT Edu workspaces via Single Sign-On (SSO). Users encountered authentication errors, including "registration_disallowed" messages, after completing their organization's SSO authentication flow. The incident lasted 25 minutes and affected the Login and ChatGPT components.

OpenAI experienced issues with account creation and login functionality affecting ChatGPT and their login system. Users were unable to create new accounts or log into existing ones across all platforms. The incident lasted 1.1 hours while OpenAI worked on implementing a mitigation to resolve the authentication problems.

OpenAI experienced increased error rates affecting the Responses API (specifically requests using the store: true parameter) and a portion of Sora traffic for 2 hours. The incident was resolved after the team identified the root cause and implemented a mitigation, with error rates declining and normal service recovering. The issue was monitored to ensure full recovery after the fix was applied.

OpenAI experienced a 19.8-hour service incident where users were unable to download or preview project files across ChatGPT, APIs, and file-related services. The issue affected file uploads and file management functionality, requiring an extended mitigation implementation period. The incident was classified as minor severity and was eventually resolved after nearly 20 hours of ongoing mitigation efforts.

OpenAI's Realtime API SIP endpoints experienced elevated error rates for 2.9 hours, affecting users' ability to access these API services. The incident began around midnight Pacific time and impacted both the general APIs and Realtime components. OpenAI identified the root cause and applied a mitigation, then monitored the service recovery.

OpenAI experienced a 1.8-hour service incident affecting sync app creation and existing app synchronization, with users encountering elevated errors including inability to send direct messages with "You cannot send messages to this room" errors. The engineering team identified the root cause and successfully applied a mitigation to resolve the issues. The service was restored to normal operation after monitoring confirmed the recovery.

OpenAI experienced a 7-hour service incident where customers encountered errors when creating and updating projects, as well as uploading files across ChatGPT and API services. The company identified the root cause, implemented mitigations, and gradually restored service functionality while monitoring for additional impacted services.

OpenAI experienced an issue where pinned chats failed to load on web browsers for ChatGPT users. The problem affected ChatGPT's web interface and persisted for 9.2 hours before a mitigation was applied and the service was restored to normal operation.

OpenAI's Sora video generation service experienced elevated error rates for 3.5 hours, affecting both the API and Sora app, with users unable to process video generation requests. The engineering team implemented mitigations that improved success rates, though high error rates initially persisted. Services eventually recovered after additional mitigations were applied and the team continued monitoring for full resolution.

OpenAI experienced high error rates affecting ChatGPT 5.4 Pro and web services for 27 minutes on March 17, 2026. The incident was classified as minor and the team investigated the issue with the affected components. The service disruption lasted from approximately 6:16 PM UTC until resolution.