Add the Gazzetta del Sud as a source
ChatGPT is down all over the world, including Italy. Since late morning today, Saturday 25 July, users are unable to load conversations, remain stuck on waiting screens or are kicked out of the session. OpenAI acknowledged the outage on its status page with an incident classified as Elevated error rateswhich involves ChatGPT, the developer APIs, and Codex, the programming platform, together. At 12.02pm Italian time the company announced that it had applied mitigation and was monitoring recovery: at the time of writing the incident remains open and performance is listed as degraded.
What doesn’t work
The reported symptoms are homogeneous throughout the world. The sidebar with the chat history remains stuck on the loading animation, old conversations do not open, sending a message returns errors related to the excessive number of simultaneous requests. Several users also report that they have been disconnected in the middle of a conversation and are unable to log back into their account: a substantial part of the reports points to the level of authentication and login rather than the actual model.
The problem isn’t just with the site. Android and iOS apps are also experiencing the same crashes, and both free and paid accounts are affected, including business plans and developers working via API. On the status page, 15 ChatGPT components, 12 API components and 4 Codex components are affected.
The chronology of the accident
The first reports on Downdetector spiked just before 11am Italian time. The portal has recorded a rapid growth in reports, which went from around two thousand to three thousand in the space of a few minutes on a global level: around 80 percent concern the general functioning of the application, 9 percent mobile apps and 7 percent Codex.
At 11.17am Italian OpenAI opened the incident on the official status page with the formula «We are investigating the issue for the listed services». Just under an hour later, at 12:02, the second update: «We have applied the mitigation and are monitoring the recovery». The company has not released a technical cause or an estimate on how long it will take for a full recovery.
The second stop in three days
This is not an isolated incident. On Thursday 23 July, ChatGPT, Codex and the API had already ended up in a prolonged outage, with a series of incidents open in the afternoon and evening and a resolution arriving only the following day, after almost twenty-four hours in total. The closeness of the two episodes has fueled users’ questions on social media about the continuity of the service, in a year in which interruptions of generative artificial intelligence platforms have been repeated with a certain frequency.
Today’s reports come from the United States, Europe, India, Japan and Australia. In Italy the discomfort is concrete for those who use the chatbot as a work or study tool: summaries, translations, revision of texts and automations connected to the API have stopped together.
What to do while the service is unstable
There are no user-side solutions: when the error is on OpenAI’s servers, no operation on your device can resolve it. Before taking the down for granted, it is still advisable to refresh the page, close and reopen the app completely and log in again instead of simply refreshing it, because in the partial recovery phases the service tends to become available intermittently again. For the rest, the only reliable reference is the official status page, which is updated at each stage of the incident.