‘AI timeout’: ChatGPT faces global outage as users report errors and slow responses

MUMBAI: Talk about a case of AI taking a timeout. OpenAI’s ChatGPT faced a partial outage on Monday, with users across the world reporting network errors, slow responses and difficulties starting or continuing tasks. OpenAI confirmed elevated latency affecting some services and said it was working on a mitigation.

The disruption appeared to be concentrated around ChatGPT Work, with OpenAI initially saying Plus users were particularly affected and that Work mode was unavailable. A later update said users across multiple subscription plans could be unable to start or continue tasks in ChatGPT Work.

The incident did not mean that every ChatGPT user was locked out. OpenAI’s service metrics cover different subscription tiers, models and error types, so the impact varied depending on the plan and feature being used.

User complaints also surged on Downdetector, with ChatGPT accounting for most of the reported problems during the disruption.

OpenAI’s status page listed the incident as a partial outage and said it had identified elevated latency in the affected services. The company said it was implementing a mitigation and monitoring the situation, but did not initially provide a specific timeline for full recovery.

The outage was focused on ChatGPT rather than OpenAI’s entire service ecosystem. The company’s API and Codex services were listed as operational at the time, meaning developers using OpenAI’s infrastructure could continue working even as some ChatGPT users faced problems.

The distinction is important for businesses that access OpenAI products through different channels. An issue with the ChatGPT interface does not automatically mean that applications built on the company’s API will be affected.

Monday’s outage also comes shortly after another disruption involving ChatGPT Work and Workspace Agents.

On August 27, OpenAI reported elevated error rates affecting those workplace features across web and mobile. The company later applied a mitigation and said the affected services had fully recovered.

The repeated issues come as companies increasingly use ChatGPT for coding, research, writing, analysis and other workplace tasks, making even short interruptions more noticeable.

OpenAI’s status dashboard showed ChatGPT with 99.67 per cent uptime between June and August. Its APIs recorded 99.94 per cent uptime, while Codex stood at 99.98 per cent over the same period.

These figures are aggregated across subscription tiers, models and different types of errors, so they do not necessarily reflect the experience of any particular user.

For now, users experiencing network errors or unusually slow responses can check OpenAI’s status page for updates as the company continues to roll out its fix. With ChatGPT increasingly woven into everyday work, even a brief AI timeout can quickly become a very real productivity problem.