OpenAI's ChatGPT experienced a widespread outage that disrupted access for users across multiple regions, temporarily preventing many from sending messages or loading conversations. The issue triggered a sharp spike in outage reports on tracking platforms such as Downdetector, while social media saw users posting complaints under the hashtag #ChatGPTDown.
According to OpenAI, the disruption was caused by a routing or backend configuration issue, not a security breach. The company said its engineering teams quickly identified the problem and implemented corrective measures. Affected services have since begun recovering, though OpenAI warned that some users may continue to experience slow responses or intermittent glitches as platform traffic stabilizes.
Reports from India suggested the outage was uneven. While some users were able to access ChatGPT without issues, others in the same regions encountered repeated error messages.
Outage data showed the highest concentration of complaints in the United States, Europe, and parts of South Asia, indicating the problem originated in specific infrastructure segments rather than a complete global shutdown.
OpenAI's status page confirmed that several related services were briefly impacted, including ChatGPT conversations, login, file uploads, search, voice mode, image generation, ChatGPT Atlas, Deep Research, Agent features, Connectors/Apps, and Codex.
The outage occurred shortly after OpenAI rolled out its Codex app for macOS on February 2. CEO Sam Altman revealed that the app saw strong early adoption, surpassing 200,000 downloads on its first day. A day later, Apple released an update to Xcode, introducing agent-based coding support for tools such as Codex and Claude.
While tech outlet 9to5Mac noted that there is no direct evidence linking these launches to the ChatGPT outage, the sudden surge in Codex-related activity may have placed unexpected strain on OpenAI's systems. However, the company has not officially confirmed this as the cause.
OpenAI has stated that a more detailed explanation may follow. Until then, users are advised to expect normal functionality to return gradually as systems fully stabilize.
