OpenAI's Status Page Logged 60+ Separate Incidents in Three Months
Published · updated · curated by AI Is Going Just Great
Source: status.openai.com ↗
All impacted services have now fully recovered.
Between June 17 and September 16, 2026, OpenAI's public status page recorded more than 60 distinct incidents across ChatGPT, the API, Codex, Sora, image generation, voice mode, file uploads, login systems, and enterprise workspaces. Affected tiers ranged from Free to Enterprise, and regional outages hit Europe, APAC, and FedRAMP environments separately. On August 20 alone, the main chatgpt.com domain went fully down, taking signups and logins with it.
No single incident dominated, which is arguably the point: the drumbeat was relentless. July 27 produced four separate status entries in one day. June 26 included a notice that a "small number of users" had their subscriptions incorrectly cancelled. August 3 required Codex remote-control users to re-pair their mobile devices after a resolution. Every entry ended with the same line: "All impacted services have now fully recovered."