OpenAI Applies Mitigation as ChatGPT and Codex Errors Hit Core Services

OpenAI has applied a mitigation for elevated errors across ChatGPT and Codex, moving a disruption that touched login, conversations and specialized tools into recovery monitoring. The company still listed the incident as degraded performance.

The disruption began at approximately 10:58 a.m. Eastern time on September 3. It affected ChatGPT conversations and login, while Codex services were also affected. The breadth matters because the incident was not confined to the core chat experience: it also reached functions used for information retrieval, file work, voice interaction and longer-running tasks.

OpenAI’s status incident counted 15 affected ChatGPT components and four affected Codex components. Those component counts describe the services marked in the incident, rather than a measure of how many people encountered errors.

Services named as affected

• ChatGPT Search, file uploads, Voice, GPTs and image generation were among the affected functions.
• Deep Research, Agent and ChatGPT Work were also named as affected services.
• Codex joined ChatGPT’s consumer and workplace functions in the disruption.

For the displayed period, OpenAI’s status page showed aggregate uptime of 99.62% for ChatGPT and 99.98% for Codex. The company says those availability figures combine subscription tiers, models and error types; individual availability may vary with the tier, model and API features in use.

OpenAI had not published a cause for the disruption at the time of the report. Its last status update said recovery monitoring was ongoing, leaving users without an explanation of what produced the elevated errors or whether the mitigation had fully restored affected functions.