ChatGPT and Major Web Services Offline as Cloudflare Suffers Global Network Failure

OpenAI’s ChatGPT and a vast network of internet applications went offline on Tuesday, leaving millions of users worldwide without access to critical digital tools. The outage has been traced to a widespread infrastructure failure at Cloudflare, the web performance and security giant that powers a significant portion of the internet. The disruption began shortly before midday UTC and has persisted into the afternoon, with engineers currently working to fully restore application services.

The first signs of trouble emerged at approximately 11:48 UTC, when Cloudflare reported it was investigating an internal service degradation. While the initial alert suggested that only some services might be intermittently impacted, the issue quickly escalated. Within minutes, users attempting to log into ChatGPT and other Cloudflare-protected platforms were met with connection timeouts and error pages.

By 12:21 UTC, Cloudflare noted that while some services appeared to be recovering, customers were still observing higher-than-normal error rates. As part of the remediation strategy, the company took specific tactical steps to stabilize the network. At 13:04 UTC, Cloudflare disabled WARP access in London, causing users in that region to see immediate connection failures. However, this measure appears to have been part of a successful identification process.

At 13:09 UTC, Cloudflare announced that the root issue had been identified and a fix was being implemented. By 13:13 UTC, critical infrastructure components including Cloudflare Access and WARP had recovered, with error levels returning to pre-incident rates. Consequently, WARP access was re-enabled for users in London.

Despite the recovery of these access protocols, the outage is not yet over for many end-users. As of the latest update at 13:58 UTC, Cloudflare confirmed that they are continuing to work specifically on restoring service for application services customers. This ongoing instability in the application layer is why platforms like ChatGPT remain inaccessible or unstable for the general public, even as the underlying network begins to heal.

Coinciding with this unexpected outage, Cloudflare was also conducting scheduled maintenance in data centers located in Santiago, Tahiti, and Atlanta. While these maintenance windows were planned to cause only minor latency or re-routing in specific regions, the broader internal service degradation has overshadowed these localized operations.

For now, businesses and individuals reliant on ChatGPT and other affected services remain in a holding pattern. Cloudflare has assured customers that their teams are focused on a complete restoration of services, though no definitive time for a total resolution has been provided.