Blocked logins, prompts, conversations, APIs for users across the globe raise concerns about reliability, recoverability and user over-reliance on the service.

Marking the latest in a series of reliability issues affecting OpenAI’s services over the past week, users of ChatGPT experienced a widespread outage on 25 July 2026, that disrupted access across multiple regions.

The outage saw many users failing to log in, submit prompts, or retrieve prior conversations through both the web interface and mobile applications, while developers have also reported problems with API access.

Data from outage-monitoring service Downdetector showed a rapid surge in user complaints during the disruption:

  • Reports in India alone exceeded 1,500 and continued climbing as the incident unfolded.
  • Multiple outlets confirmed that OpenAI had acknowledged the issue as a global outage, with users unable to load chat histories or initiate new sessions.
  • Outage reports spanned the United States, Europe, India, Japan, Australia, and the United Arab Emirates.
  • Breakdowns in core ChatGPT functionality accounted for the majority of reported problems, representing roughly 80% of incidents logged on Downdetector.
  • The remaining issues were linked to mobile app instability and API connectivity failures. OpenAI’s Codex platform, used for coding-related tasks, was also affected.

On its official status page, OpenAI has classified the situation as “elevated error rates” and stated that engineering teams were actively investigating the cause.

This latest outage follows several earlier recent incidents, raising broader concerns about service stability. On 21 July 2026, users reported problems with image generation features alongside general performance issues. Another significant outage had occurred on 23 July 2026, impacting ChatGPT, Codex, and the firm’s APIs. For several hours users experienced elevated error rates, and although mitigation measures were implemented, full recovery was not achieved until nearly a day later.

The clustering of these incidents has drawn attention to operational resilience as demand for AI services continues to scale. The firm has not identified a single root cause linking the recent incidents. In prior outages earlier in 2026, it had attributed disruptions to factors such as configuration changes, failures involving upstream service providers, and errors introduced during system deployments. The absence of a clear explanation for the latest sequence of outages may further intensify questions around infrastructure robustness and change management practices as reliance on AI platforms continues to grow.