OpenAI marked an October 22 Dots incident resolved after reporting failures when users tried to send messages to their agents. Its investigation update said Enterprise customers accounted for most of the impact, rather than describing the problem as exclusive to Enterprise. The final update says affected services recovered. This is a service-recovery report, not a new feature announcement or a claim that an outage remains active.
The public incident page identifies Dots as the affected ChatGPT component and records investigation, identification and resolution. It does not explain the underlying cause, quantify affected accounts or establish whether any interrupted task resumed automatically. That leaves an important distinction for people managing ongoing work: a recovered messaging service is not, by itself, confirmation that a particular assignment finished.
Editorial take: If an important request was sent during the disruption, check the conversation and its expected output before repeating it. Ask the agent to identify what completed, what remains pending and whether anything needs another instruction. For work involving an external message or change, inspect the destination first so that recovery does not become an accidental duplicate. Keep the original request and any confirmation handy if a missing result needs further investigation. We have not independently tested affected accounts.