OpenAI's acknowledgment used the singular: it was "investigating the issue for the listed services" [3]. One declared fault against fifteen listed components [9] points upstream, to something shared by Login and the Compliance API.
Login is the entry in that list that changes the shape of the incident. When authentication is in the failure set, there is no degraded mode: a component still serving requests is unreachable to anyone who does not already hold a live session [10]. Users saw conversations fail to load and errors on sending messages [8], which is what that looks like from the client side.
What the public record does not contain is a cause, a restoration time, a count of affected accounts, and any statement about the developer API platform. The named components are ChatGPT surfaces plus Codex [13]. OpenAI was still investigating elevated errors when the report went out [5], so the only duration anyone can quote is open-ended from 10:58 AM ET [11].
The account itself is thin by construction. Both available versions are the same BleepingComputer article, so this amounts to one publisher reading a vendor status page [12]. That establishes which components failed together and when it started; it does not establish blast radius.
On timing: the outage landed ahead of Astra, which OpenAI has confirmed will ship soon [6]. OpenAI has not said the two are connected, and the judgment that they are probably unconnected is BleepingComputer's, not the status page's [7]. The material does not show a launch-window cause.
The component list is the artifact worth keeping. It records which workflows fail in the same minute, and it puts a desktop Codex session and a consumer chat session in one failure domain [14]. Teams that wired either one into daily work now have a documented answer to how far the outage reaches, which is further than the chat window.
