A growing body of research is converging on a troubling conclusion for the open web: AI-generated search answers are systematically dismantling the economic model that has kept independent publishing alive for two decades.
Two academic papers and new usage data from Pew Research paint a picture of accelerating decline. AI Overviews, the summarized answers that Google and other search engines now place above traditional links, are reducing organic clicks to publisher sites by approximately 40 percent while increasing zero-click searches by 35 percent, according to a field experiment by researchers at the Indian School of Business and Carnegie Mellon.
The broken bargain
The core problem is structural. Publishing on the web has worked on a simple bargain: publishers produce content, search engines send readers to that content, and the visits generate revenue through advertising, subscriptions, or affiliate links. AI search answers break that loop by extracting the information and keeping the user inside the search interface.
Alex Chan of Harvard Business School, in a paper titled “AI and the Collapse of the World Wide Web,” describes this as a loss that goes beyond immediate revenue. When an AI answer diverts a visit, the publisher loses not just the page view but the durable signals that sustain its business over time: subscribers, repeat readers, backlinks, bookmarks, search authority, and reputation.
“When an AI platform diverts the revenue and measurement events without replacing them, costly human information may fall below replacement,” Chan writes.
More than just lost clicks
The damage is compounded by the difficulty of distinguishing human-authored content from AI-generated material. Chan argues that new market mechanisms, provenance tracking, diversity pricing, exploration credits, and informative audits, are needed to restore the ecosystem. Simple fixes like requiring AI companies to pay for content or banning AI answers altogether are impractical, he says.
Pew Research data from July 2025 had already shown that users click links less often when AI summaries appear in search results. The new research quantifies the effect and confirms it is accelerating as AI summaries become more comprehensive.
Self-reinforcing decline
The structural risk is that the process becomes self-reinforcing. As AI search reduces traffic to publisher sites, publishers produce less original content. With less original content to draw from, AI answers rely more heavily on recycled or AI-generated material, reducing overall quality. Users get worse information, and the incentive to produce costly human journalism, analysis, and investigation weakens further.
Cloudflare’s recent move to block mixed-use AI crawlers from ad-supported pages is one industry response to the same underlying tension. But the research suggests that technical blocking alone is not enough, the web needs new economic mechanisms for the AI era, and those mechanisms do not yet exist.