Topline
OpenAI on Tuesday said a fix has been applied to a “partial outage” disrupting functionality for ChatGPT, though the company said the AI-powered chatbot was still impacted by “degraded performance,” as some users still report issues on DownDetector.com.
OpenAI said it has “identified the root cause” of the issue and is working on a fix.
Copyright 2023 The Associated Press. All rights reservedKey Facts
OpenAI reported issues with the company’s application programming interfaces, ChatGPT and its AI-powered image-generating service Sora starting around 3 a.m. EDT, saying users are “experiencing elevated error rates and latency” across several services.
More than 1,900 users reported issues across OpenAI’s service on DownDetector.com on Tuesday morning, with 92% of reports indicating issues with ChatGPT, though the number of outage reports dropped below 400 just before noon.
At around 9 a.m., OpenAI said it “identified the root cause of the issue” and has been “working on implementing a mitigation” for the outages.
Performance for API and ChatGPT improved at around 11:42 a.m., according to OpenAI, which upgraded issues with ChatGPT from a “partial outage” to “degraded performance” as a “mitigation” was implemented at 1:20 p.m.
OpenAI did not immediately respond to a request for comment.
Big Number
More than 500 million. That’s the estimated number of active weekly users ChatGPT had as of March 31, according to OpenAI.
Key Background
ChatGPT’s latest outage follows other issues with OpenAI’s chatbot in recent months. A similar outage was reported among ChatGPT users in March, during which OpenAI reported “increased error rates” with ChatGPT, Sora and other products. That outage peaked at just over 1,100 outage reports and was resolved within hours. Another ChatGPT outage came in January, after OpenAI said it was experiencing “elevated error rates” in the API, though the issue was also resolved within a few hours. Widespread issues with ChatGPT were reported in December, when more than 50,000 outages were reported on DownDetector.com. The outage, first reported in the early afternoon, was resolved after more than six hours.
Further Reading
Read the full article here
