- Sources: OpenAI status page, HN 49046049
- Summary: OpenAI opened an incident titled elevated errors affecting ChatGPT conversations at 2026-07-25 22:09 UTC, covering intermittent errors that prevented some users from loading or continuing ChatGPT conversations, dated impact from about 13:00 PT, identified the source at 23:16 UTC, and applied a mitigation at 23:57 UTC. It was read at this run as still open. The status page carries a full-outage marker and header text stating that OpenAI is currently experiencing issues, with the incident in the Monitoring state, a note that mitigation has been implemented and recovery of ChatGPT conversations is being monitored, and a duration label of 15 hours. No root cause is published. Separate Ask HN threads titled ChatGPT Is Down and Codex Is Down appeared the same day, so user-visible impact extends past the web product, but the status page text readable here does not name Codex as an affected component. The incident permalink returns a JavaScript shell to this environment, so the update timeline could not be read directly and the details above come from the root status page, whose content changes once the incident closes.
- Why it matters: A mitigation applied at 23:57 UTC that still leaves the incident open 15 hours later means the fix is unconfirmed, so retry and fallback paths against the ChatGPT surface stay load-bearing.
- Follow-up: Watch for the incident closing, for a root cause, and for whether Codex is named as an affected component in the final update.
send feedback on this story