A major part of the internet briefly went offline on Tuesday after a sudden systems failure at Cloudflare disrupted access to several high-traffic platforms, including ChatGPT, X, and Downdetector. The disruption lasted long enough to cause global confusion, but Cloudflare has now confirmed that the issue came from an internal technical error—not a cyberattack.
Cloudflare CEO Matthew Prince shared a detailed explanation in a late-night blog post, saying the outage began inside the company’s Bot Management system. This system normally helps protect websites from harmful traffic and manages how automated crawlers interact with Cloudflare’s network. Since nearly 20 percent of all global web traffic runs through Cloudflare, even a small failure can spread quickly—and that is exactly what happened.
According to the company, one of the databases controlling bot-related permissions experienced a faulty change. This triggered a malfunction in the machine learning model responsible for assigning bot scores to incoming traffic. Cloudflare said that because of a change in how ClickHouse queries collected data, the system produced many duplicate data entries. This made the configuration file grow far beyond its memory limits.
When that file became too large, Cloudflare’s primary proxy system stopped working correctly. This proxy handles traffic for customers who depend on bot-related rules. As a result, genuine users trying to access websites were flagged as bots and blocked. Platforms using other Cloudflare services continued running normally, but many major sites relying on bot filtering were knocked offline.
The incident happened as Cloudflare has been tightening controls around automated crawlers, especially those used for scraping content to train AI models. Although its new AI-focused systems were not involved in this outage, the company admitted the event highlights the risks of relying on shared infrastructure—a challenge already seen in recent outages at Microsoft Azure and Amazon Web Services.
Cloudflare says it is now taking immediate steps to prevent this type of failure from happening again. While the company did not list every technical fix, it confirmed that improvements are being applied to configuration handling, system checks, and safeguards around critical updates.
The outage serves as a reminder of how interconnected the online world has become. A single error inside one service provider can quickly impact millions of users worldwide. However, Cloudflare insists the issue has been resolved and all systems are now operating normally.
