OpenAI's ChatGPT recently experienced a global service outage, with users unable to log in, register accounts, load historical chat records, or send messages normally. The affected areas include the United States, Europe, and other regions.

The incident began around 8 p.m. on August 19, U.S. Eastern Time. Affected users who opened ChatGPT might find the page stuck on the sidebar loading animation and see error messages such as "Too many concurrent requests," preventing them from sending messages. Meanwhile, the new user registration and account login functions on chatgpt.com also encountered issues, with some users unable to access previously saved chat records.

ChatGPT

Aside from the main ChatGPT service, OpenAI's coding platform Codex was also affected by this outage. OpenAI later confirmed the issue through its official status page, stating that it had identified some users experiencing login problems and was taking measures to mitigate and restore the service.

This incident also affected OpenAI API services. According to information on the OpenAI status page, multiple API endpoints experienced anomalies, affecting developers' ability to call related AI services.

As of around 8:15 p.m. U.S. Eastern Time, OpenAI confirmed that the failure was still ongoing and stated that the engineering team was working on resolving the issue. This large-scale service disruption once again highlights the importance of the stability of AI infrastructure. As ChatGPT and other generative AI tools gradually become core services for both individual users and enterprise development processes, platform availability, fault response, and system resilience will become key indicators of competition among AI service providers.