A widespread internet disruption on Tuesday temporarily slowed or blocked access to several major digital services, including ChatGPT, X, Spotify, Canva, and Claude, after network infrastructure provider Cloudflare suffered a large-scale outage. The incident affected millions of users and businesses worldwide, highlighting how dependent the modern web is on a small number of network operators.
Cloudflare Chief Technology Officer Dane Knecht described the problem as a “latent bug” — a flaw that typically remains hidden until triggered under specific conditions. In this case, a database began generating incorrect entries in a configuration file used for Cloudflare’s bot protection system. The file unexpectedly doubled in size, and as it propagated across Cloudflare’s global network, thousands of servers tried to load more data than they were designed to handle. This caused routing systems to fail, triggering widespread service disruptions.
The outage exhibited an unusual “dropping and returning” pattern. Some servers continued producing valid files while others generated faulty ones, causing the system to intermittently recover before failing again minutes later. Initially, engineers suspected a massive DDoS attack due to the fluctuating failures. Once the issue was diagnosed, Cloudflare stopped the faulty file generation, manually pushed a stable configuration, and restarted affected services. Core traffic resumed by 14:30 UTC, with all systems returning to normal by 17:06 UTC.
Cloudflare apologised for the impact in a post explaining the failure. “We know we let you down today,” the company said, noting that the disruption was particularly serious given Cloudflare’s central role in global internet infrastructure. The company promised a detailed incident report and measures to prevent similar failures in the future.
Cloudflare provides routing, traffic protection, and other network services for roughly 20% of the world’s websites, with data centres in 330 cities and connections to over 13,000 networks, including major cloud providers and telecom operators. The outage underscores the critical role of a few infrastructure companies in keeping the internet running. Ironically, the failure came from an internal error in a system designed to prevent massive traffic spikes, the very kind of event Cloudflare protects clients against.
While internet traffic has returned to normal, some users may continue to experience issues accessing the Cloudflare dashboard as residual problems are resolved.
