The social media site X was among those affected by the problem

Paris (AFP) - Major websites including social network X and AI chatbot ChatGPT were disrupted on Tuesday after US online services provider Cloudflare said it had been affected by a “latent bug”.

Web monitor Downdetector recorded disruptions for users of X, video game “League of Legends” and some services from Google and OpenAI, the creator of ChatGPT.

Cloudflare, which specialises in online security and says it manages some 20 percent of global internet traffic, saw its share price slump 1.5 percent in early trading.

“Earlier today we failed our customers and the broader internet when a problem in Cloudflare network impacted large amounts of traffic that rely on us,” chief technology officer Dane Knecht wrote on X, adding that the problem had since been resolved.

“In short, a latent bug in a service underpinning our bot mitigation capability started to crash after a routine configuration change we made.”

The company said earlier there had been “a spike in unusual traffic” to one of its services.

The outage was reminiscent of those that hit Amazon (AWS) and Microsoft cloud services last month, disrupting some online services for video games, businesses and transport firms.

“This incident, as with the recent outage at AWS, shows how reliant some very important internet-based services are on a relatively few major players,” said Alan Woodward, professor of cybersecurity at the University of Surrey in England.

“It’s a double-edged sword as these service providers need to be large to provide the scale and global reach required by big brands. But when they fail the impact can be significant.”

The CloudFlare outage is the latest in a trend of infrastructure providers going offline and taking swaths of the internet with them, according to Emarketer analyst Jacob Bourne.

While Cloudflare carries about a fifth of global internet traffic, it touches a third of the world’s top websites - powering retailers like Shopify and AI providers like OpenAI and Anthropic, as well as smartphone apps and streaming services, the analyst noted.

“We’re seeing outages happen more frequently, and they’re taking longer to fix,” Bourne said.

“That’s a symptom of strained infrastructure: increased AI load, streaming demand, and aging capacity all pushing systems past the edge.”

The latest outage spotlights the importance of seemingly mundane internet infrastructure to the “AI Revolution”, according to Cornell University Tech Policy Institute director Sarah Kreps.

“The issue exposes the reality that this multi-billion, even trillion dollar investment in AI is only as reliable as its least scrutinized third party infrastructure,” Kreps said.

Knecht put out word on X that by day’s end Cloudflare will share details of what went wrong along with what the company plans to do to make sure it doesn’t happen again.