Inside Web Publishers’ Quest to Stamp Out AI Bots Posing as Humans

In recent months, some websites have become increasingly concerned about an influx of traffic from what they suspect are artificial intelligence chatbots, which can drive up those sites’ costs and swipe their content. But separating the bots from legitimate human visitors is no easy task.

Several months ago, for example, a staffer at cybersecurity firm DataDome prompted xAI’s chatbot Grok to visit a website DataDome runs to analyze security threats. The website rarely gets much web traffic, but after the Grok request, it got a dozen visitors within the space of five to 10 seconds. One was clearly a bot, but others visited from a Firefox browser or an iPhone, as a human might.