Cloudflare explains bug behind global internet slowdown that took X, ChatGPT and others down

Cloudflare

A major part of the internet stumbled on Tuesday morning after a widespread Cloudflare outage took down or severely impacted services including ChatGPT, Claude, Spotify, X and several other popular platforms. The disruption began around 8 a.m. ET, when users across the world reported that apps and websites were either refusing to load or throwing server errors.

Cloudflare, which powers a significant portion of the internet’s infrastructure, said it had identified the issue early in the morning and began rolling out a fix soon after. By mid-morning, the company said the “incident is now resolved” but added that it would continue monitoring systems to ensure everything returned to normal.

In an unusually candid explanation on X, Cloudflare CTO Dane Knecht apologised and said the outage stemmed from a latent bug inside a core service that powers the company’s bot mitigation systems. A routine configuration change triggered the previously undetected bug, causing a cascading failure across Cloudflare’s network. “This was not an attack,” Knecht clarified, acknowledging that the company had “failed our customers and the broader internet.”
Knecht promised a full post-mortem in the coming hours and said Cloudflare is already working to ensure the same issue doesn’t repeat. While most major services have recovered, the company noted that some users may still face problems logging into the Cloudflare dashboard as it continues deploying fixes.
The incident arrives less than a month after a similar outage at Amazon Web Services and is yet another reminder of how dependent the modern web is on a small set of infrastructure providers. Cloudflare alone powers 20 percent of all websites globally, with 330 data centres and connections to 13,000 networks worldwide.