AI & Tech · San Francisco
OpenAI confirms ChatGPT down globally: thousands affected, login errors and blank chat histories reported
On Saturday, July 25, 2026, ChatGPT experienced a global outage that prevented millions of users from logging in, accessing conversations, or generating responses. OpenAI confirmed the disruption and said it was working on a fix.
In brief
DownDetector recorded more than 3,000 complaints on Saturday; 79% concerned ChatGPT, 9% the mobile app, and 8% the Codex platform.
Users in North America, Europe, and Asia reported login failures, empty chat histories, and server errors beginning around 13:00 EEST on July 25.
OpenAI confirmed the high error rate, applied remediation measures, and said it was monitoring recovery. Early reports showed no evidence of a malicious intrusion.
The outage extended beyond the consumer chatbot to OpenAI's API for developers and its Codex code-generation tool, disrupting third-party services.
What happened
On Saturday, July 25, 2026, around 13:00 Romania time, users across the globe began reporting they could not log into ChatGPT, view past conversations, or generate new responses. The platform displayed errors including "primaryapi_server_error" and "500 Internal Server Error," or remained stuck on a loading screen. The outage also extended to OpenAI's API services and its Codex coding assistant, leaving third-party developers and business applications without access to the company's AI models.
Global reach
The disruption was not isolated to any region. Reports poured in from the United States, India, and across Europe: the United Kingdom, Germany, Italy, Spain, Austria, Greece, and Croatia all saw spikes in complaints. Users in Singapore, Malaysia, the Philippines, Indonesia, and Australia also faced the same issues. Over 3,000 reports were filed on DownDetector during the day, with 79% linked to ChatGPT, 9% to the mobile app, and 8% to Codex. Google Trends showed a surge for the query "Is ChatGPT not working?" as people sought confirmation of the outage's breadth.
- ChatGPT
- 79 %
- Mobile App
- 9 %
- Codex
- 8 %
Data from the article text
OpenAI's response
OpenAI acknowledged the incident on its official status page, stating it was "investigating the issue" and experiencing "a high rate of errors." The company later added that it had applied remediation measures and was monitoring recovery. No root cause was officially released, though possible explanations included an internal system error, a configuration problem, or an infrastructure malfunction. The Italian newspaper La Stampa reported that there were no early indications of a cyberattack.
“We have applied remediation measures and are monitoring the recovery process.”
User reaction
While engineers worked on a fix, users turned to social media platforms like X, Reddit, and TikTok. Many posted memes about having checked their internet connection or restarted their devices before realizing the outage was global. The humorous response reflected how embedded the AI chatbot has become in daily routines.
Technical scope
The failure affected not only the consumer-facing ChatGPT but also the OpenAI API, which powers numerous commercial products, and Codex, a tool for generating and editing code. The incident highlighted the dependency of both individuals and enterprises on OpenAI's infrastructure. As of the latest updates on Saturday, no timeline for full restoration had been provided.