In the early morning of December 27th, Beijing time, OpenAI's AI service suffered a large-scale interruption, affecting ChatGPT, Sora and some API calls. The outage caused widespread concern as users were unable to use these services. Although OpenAI announced at 6:05 a.m. that ChatGPT has partially resumed service, the chat history loading function is still limited, and the full recovery time has not yet been determined. OpenAI did not elaborate on the specific cause of the outage, only calling it "an issue caused by the upstream provider." Microsoft Azure also reported power issues in its data centers, suggesting a possible connection between the two.
At 6:05 a.m., OpenAI announced that ChatGPT had partially resumed service, but the chat history loading function was still limited, and the time for full recovery has not yet been determined.
Regarding the specific cause of the service interruption, OpenAI only stated that it was "caused by the upstream provider." Microsoft Azure, OpenAI's exclusive cloud service provider, also reported a "power supply" problem in its data center on the same day.
This is not the first time that the ChatGPT service has experienced interruptions. Similar service interruptions have occurred many times in the past few months.
The OpenAI service outage once again highlighted the stability issues of large-scale language model services and reminded us to remain vigilant about the risks of relying on such services. In the future, improving service stability and improving emergency plans will be what AI companies need to focus on to ensure user experience and data security.