ChatGPT Goes Down Again: Users Hit by Strange Lockouts Worldwide
ChatGPT users report sudden logouts and unusual activity errors as OpenAI confirms a routing issue behind a second major outage in two days.
image for illustrative purpose
On Tuesday, OpenAI’s main chatbot, ChatGPT, suffered a massive outage once again, making this the second consecutive day of malfunction for users trying to access the platform. The outage reports on DownDetector, which counts the real-time user complaints worldwide, saw a surge of over 12,500.
OpenAI mentioned that during this outage, the platform experienced a high level of error activity. The company subsequently diagnosed the problem as an internal routing issue and stated that service was back on after a fix was implemented by the engineers.
The complaints included about 93% of the users who were dealing with ChatGPT. The other tools of OpenAI seemed to function without any issues. Countless users reported that they were logged out of their accounts without any warning and could not access past chats. A good number of them also got an alert saying, "Your device has been showing unusual activity. Please try again later.”
The reports indicated that the web interface was the least stable while the Mac desktop app was operational only sometimes.
The data from DownDetector indicated that almost 3,000 users lost access during the worst period of outage.
The whole thing was happening during a time when OpenAI had set a goal for the development team to make the service more consistent and reliable. Recently, CEO Sam Altman described what he called a "code red" aimed at improving system reliability after users faced hiccups repeatedly, which was the reason behind the issuance of his "code red".
Moreover, shortly before the outage, OpenAI announced that Mixpanel, its analytics partner, had been breached and names, email IDs, and API-related information tied to some users had been accessed without authorization.
Over the course of two days, there were numerous reports of widespread issues but OpenAI’s public status page did not show any disorder right away, which raised queries about the efficiency of internal monitoring tools among developers and users.
