If you tried to use ChatGPT today and got nothing but error messages, you weren’t the only one. Thousands of people across the globe found themselves staring at blank screens this afternoon as the world’s most popular AI chatbot hit a major snag. According to Down Detector, a site that tracks website crashes, reports of issues spiked to over 12,000 during the worst part of the outage.
OpenAI, the company behind the tool, quickly acknowledged that things were broken. They posted a status update noting that “elevated error rates” were affecting both regular ChatGPT users and developers who use their platform to build their own apps. For a few hours, the digital assistant that many people now rely on for writing emails, summarizing documents, or coding just wouldn’t cooperate.
The company worked on a fix throughout the afternoon and finally announced that they had resolved the main issue around 5:14 PM ET. While the chatbot started working for most people again, OpenAI kept one alert active for a bit longer. This specific issue affected the “fine-tuning” part of their service, which is a tool developers use to train the AI on specific sets of data. OpenAI eventually applied a fix for that as well and monitored the system to make sure the errors didn’t come back.
Strangely, ChatGPT wasn’t the only AI having a bad day. Anthropic’s Claude, a major competitor in the AI space, also went dark for a while today. Users trying to access Claude’s various models through its API ran into similar error messages that prevented them from getting any work done. Anthropic managed to get their systems back under control a bit earlier than OpenAI.
Neither company has explained exactly what caused these hiccups or if the two outages were related in any way. For many office workers, students, and programmers, the afternoon was a sudden reminder of how much we’ve come to depend on these tools in a very short amount of time. When the AI stops working, it seems a lot of productivity stops right along with it.
