Google's search deal with the web is dead — publishers now consider blocking crawlers

For two decades, the web operated on an unspoken contract: Google scraped content, built the world’s dominant search engine, and sent trillions of visits back to publishers. It was lopsided but functional. That deal, by most accounts, is now over.

The Vergecast team dug into what they’ve branded ‘Google Zero’ — a future where Google’s AI-generated answers satisfy queries without ever sending users to the source. Nilay Patel took a brief victory lap for coining the term, but the news this week suggests the phenomenon is accelerating. Reddit, one of the web’s largest repositories of human conversation, is reportedly looking for an exit from an AI training deal. Meanwhile, publishers who once lived and died by search traffic are now openly weighing whether to block Google’s crawler entirely.

David Pierce and Nilay Patel didn’t mince words: even Google’s core Search business is getting squeezed by the AI shift. The irony is thick. The company that trained its models on the open web is now building products that make visiting that web unnecessary. And the sites that supplied the data? They’re realizing they may have negotiated their own obsolescence.

Elsewhere in the episode, OpenAI’s accidental ‘hack’ of Hugging Face raised fresh questions about how much trust we should place in AI labs that move fast and break things — including other people’s platforms. The crew also lit up the Hype Desk to torch dubious AI detection tools now popping up across platforms, before touching on Ford’s integration with Apple Maps and the arrival of a new minimalist phone that, mercifully, probably doesn’t have an AI assistant.