Google takes the content, keeps the audience, and calls it search. That is the arrangement publishers have spent two decades accepting, and the remarkable thing about the current moment is not that companies like USA Today and Reuters are threatening to block Google’s crawler — it is that the traffic had to fall by half before anyone with a balance sheet said it out loud.
It is true that Google’s AI search features send traffic to publishers. Google says its AI Overviews deliver billions of clicks to the web every week, and there is no reason to doubt the raw number. The trouble is that the number is a floor, not a trend. USA Today’s organic search traffic from U.S. users fell by nearly half between June 2025 and June 2026, according to search measurement firm Semrush. Business Insider’s dropped more than 85 percent over the same period. Politico lost 23 percent, CNN about 25. People Inc. — whose CEO, Neil Vogel, now says blocking Google’s crawler is “100 percent on the table” — watched Google’s share of its traffic fall from more than half to 25 percent in two years. The clicks are real. They are also, for a growing number of publishers, insufficient to sustain the newsroom that produced the content Google is summarizing.
It is worth being precise about what Google’s AI Overviews actually do, because the public discourse has the misleading habit of treating them as a search improvement and they are, in what they actually perform, something else entirely. When a user searches for a news event or a how-to question, Google’s large language model reads the publisher’s article — the one a reporter wrote, an editor fact-checked, and a newsroom paid to produce — and generates a summary that appears above the organic search results. The user reads the summary. The user does not click through. The publisher provided the content. Google kept the audience.
That is not a search result. It is a content-repackaging operation that routes the audience to Google’s own surface and sends the publisher the remainder — which, if the Semrush numbers are representative, is approaching nothing for a growing number of outlets.
Cory Doctorow has a name for this pattern. In the first stage of what he calls enshittification, a platform is good to its users: Google sent readers to publishers, and publishers built their businesses around the traffic. In the second stage, the platform is good to its business customers — and Google’s ad-tech arrangement was, for a time, the engine that funded digital journalism at scale. In the third stage, the platform claws value back from both sides for itself. AI Overviews are the claw. The content was always the product; what changed is that Google no longer needs to send the audience to the publisher to deliver it.
The four forces that once constrained this kind of extraction — competition, regulation, interoperability, and labor — are, in this case, almost entirely absent. Google controls more than 90 percent of the search market. Publishers have no alternative distribution channel at comparable scale. The standard “two-sided market” framing treats search as a neutral intermediary serving both sides equally; what the AI-summaries shift reveals is that Google is collapsing that model into one-sided extraction, capturing both the supply of content and the demand for attention without returning adequate value to either. The regulatory environment is only now beginning to respond: the U.K. regulatory authority has required Google to allow publishers to opt out of AI features without losing their placement in traditional search results. Google has said it will test the U.K. features before rolling them out globally. The willingness to comply rather than litigate for years — as Google has in the ad-tech antitrust case — suggests the company has calculated that compliance is cheaper than the fight. That is a calculation worth tracking.
The structural problem is that Google is what economists call a monopsonist — a dominant buyer — positioned between publishers and their readers. When one company controls 90 percent of the search market, the publisher cannot walk away; it can only negotiate the terms of its own extraction. The question no single publisher can answer alone is whether a bloc of them, acting together, could force a renegotiation — the classic collective-action problem that keeps extraction stable. Each publisher fears that blocking Google alone means disappearing from search while the others remain, and that coordinated action requires a trust none of them can afford to extend to a competitor. Reddit, which struck a $60 million deal with Google in 2024 to allow its content to be used for AI training, is now reassessing that arrangement as it approaches its end. The $60 million bought Reddit the privilege of becoming the training data for the AI that now summarizes Reddit’s posts without sending users to Reddit. That is not a licensing deal. It is a racket structured as a press release.
The most revealing detail in the current discussion is not any publisher’s threat to block Google. It is Time magazine’s alternative strategy. Time’s chief operating officer, Mark Howard, described what the publication is doing: creating text-based ads invisible to human readers but designed to be picked up by AI crawlers. “There are now two audiences,” Howard said. “We’re thinking about, How do we superserve the humans when they do come, and how do we think about the bots as a secondary audience?” That sentence is worth reading twice — not as a sign that Time has abandoned journalism, but as evidence of what the monopsony structure makes rational. When the crawlers outnumber the readers and Google controls the valve between content and audience, targeting the machine is not a strategic preference; it is the profit-maximising response to an incentive structure that punishes any publisher who tries to reach humans the old-fashioned way. The indictment falls on the structure that makes this the intelligent move, not on the actor making it.
Cloudflare reports that bots account for more than half of all web traffic. When the crawlers outnumber the readers, and the publisher’s revenue strategy is to target the crawlers, the publisher is no longer in the business of journalism. It is in the business of feeding the model.
The publishers threatening to block Google’s crawler are, in effect, claiming the most basic right a content producer can assert: the right to decide who uses your work. That this right is currently theoretical — that publishers cannot reliably distinguish human readers from AI crawlers, and that blocking Google’s bot risks disappearing from search entirely — tells you how completely the platform has captured the distribution channel. USA Today’s Mike Reed called it “taking a stand.” It is also, structurally, a publisher admitting that the relationship it has depended on for two decades has become extractive enough that death by blocking might be preferable to death by summary.
The German court ruling in June and the U.K. regulatory authority’s requirement that Google let publishers opt out of AI features without penalizing their traditional search rankings establish, in principle, the right structural fix. Both decouple the extraction from the distribution. If publishers can say no to AI Overviews without losing their search placement, the monopsony’s leverage breaks, and Google has to compete for the content it has been taking for free. The test is not what Google says about compliance. The test is what happens to publisher traffic in the quarters after the opt-out mechanisms go live.
Google says it sends billions of clicks to the web every week. The publishers who produced the content those clicks are built on are laying off the reporters who wrote it. Both numbers are real. They describe the same system, running in one direction.