Google's web traffic deal is breaking down as AI search reshapes the internet

On The Vergecast episode from July 24, 2026, David Pierce and Nilay discuss how Google's foundational relationship with the web is unraveling. For years, Google collected data by indexing webpages and, in return, sent substantial traffic to those sites through search results—a deal that benefited Google far more than publishers but sustained the web's economic model. That arrangement is now breaking down. As AI search tools answer questions directly without directing users to the original sources, websites are losing the traffic that powered their business. In response, some publishers are considering blocking Google from crawling their sites entirely, a drastic move that reflects their desperation as Google traffic disappears. Reddit is also reassessing its participation in AI training data deals, suggesting major publishers are rethinking their relationship with the platforms mining their content. The shift is so significant that it is affecting Google's own Search business, which is experiencing pressure from the broader AI transition. The episode explores what the web looks like when the Google traffic deal dies—a question with no clear answer yet. Beyond the Google situation, the Vergecast also covers reporting on OpenAI's accidental intrusion into Hugging Face systems, growing concerns about the reliability of AI detection tools spreading across platforms, and other developments including an Apple payment plan, a new minimalist phone design, and Ford's partnership with Apple Maps.

For the past two decades, a fundamental compact existed between Google and the broader web: Google indexed webpages and served them in search results, directing enormous traffic back to publishers who depended on those clicks for revenue. That deal was asymmetrical—Google extracted far more value than publishers received—but it held. Now, however, the mechanics of AI search are breaking that arrangement. When large language models can synthesize answers to user questions without routing searchers to specific publisher pages, the incentive for publishers to make their content available disappears. Reddit's reconsideration of AI training agreements and publishers' threats to block Google's crawlers reflect a recognition that the old model no longer works. Even Google's own Search business faces pressure as the AI landscape shifts how users find information, creating uncertainty about what replaces the decades-old web ecosystem that relied on search traffic.