According to Downdetector, issues kicked off at around 2.45pm IST and quickly ballooned, with more than 1,100 complaints clocked by 3.02pm. A whopping 93 per cent of those grumbles were aimed directly at ChatGPT, while a sad seven per cent were about the app itself and a lonely one per cent got stuck in login limbo.

X, formerly Twitter but still just as unhelpful, lit up with screenshots of broken chats, endless login loops, server errors and baffled users watching sessions vanish into the ether. Several poor souls reported lost work and ghosted conversations, all without so much as an apology from the chatbot that forgot how to chat.

In fact many users found that ChatGPT was rather keen to blame them for the outages with errors saying their networks were at fault, or there were too many requests made by the same user. Of course, neither of these are possible if the entire system is down.

OpenAI eventually stirred itself into action, slapping a mitigation in place by 10.50pm IST. Recovery limped along with both the ChatGPT interface and its API showing signs of life. By 1.04am on 11 June, the company was crowing about full API recovery, although voice mode remained shaky.

The firm’s status page now says things are in “monitoring” mode, which usually translates as “we think it’s fixed, but who knows?” While OpenAI has yet to spill the beans on what went wrong, it insists the fix is holding and they’re keeping a close eye on it.

The glitch didn’t discriminate, hammering free and Plus users across browsers, mobile apps, and GPT API integrations. No official word on data loss, but plenty of users claimed their sessions got nuked mid-flow or responses just stopped mid-thought.

This means that all those companies who fired people and replaced them with AI bots were effectively shut down and developers who rely entirely on it for coding were either having to go back to what they did in the old days or take a long, long lunch break.