Ethics
Publishers Weigh Cutting Off Google As AI Summaries Squeeze Search Traffic
Reddit and major publishers are reportedly considering whether to restrict Google's access to their content as AI summaries reduce referral traffic and weaken the old search bargain.
By Michael C ยท
Reddit and several major publishers are reportedly considering whether to restrict Google's access to their content, a sign that AI summaries are weakening one of the internet's oldest commercial bargains. The Wall Street Journal reported, through accounts cited by multiple outlets, that Reddit, USA Today, Politico, Reuters, The Economist, and People Inc. are reassessing how much they want Google to crawl or use their material as AI-generated answers reduce referral traffic. The old exchange was simple: Google indexed pages and sent readers back. AI Overviews make that exchange less predictable because the answer can appear before the click.
The numbers being discussed are severe. Reports citing Semrush data say Google search traffic from U.S. users fell by nearly half for USA Today's national paper between June 2025 and June 2026, by 23% for Politico, by about 25% for CNN, and by more than 85% for Business Insider. Google disputes the industry's framing and says its AI features still send billions of clicks to the web. Both claims can be true at the same time if total click volume remains large while individual publishers see sharp declines in important categories.
Reddit's position is unusually complicated. It has a data licensing agreement with Google reportedly worth about $60 million a year, and it also has a separate relationship with OpenAI. Its content is valuable because human discussions answer questions in ways that formal articles often do not. But if AI summaries use that value to answer users directly, Reddit has to ask whether a flat data payment compensates for lost traffic, weaker ad inventory, and reduced control over how its communities are represented.
For news publishers, the choice is even more severe. Blocking Google's crawler can mean losing AI exposure and traditional search exposure together, depending on the tools available and Google's policies. That makes a boycott costly. A publisher can object to AI summaries but still depend on Google for audience acquisition. The power imbalance is the heart of the story. Publishers supply the material that makes search useful, yet they often have limited leverage over how the search page evolves.
Google has said website owners can control aspects of how their content is used, and reports say the company is preparing ways for publishers to opt out of AI features without disappearing from ordinary search in some contexts. That kind of unbundling would matter. The current fear is that publishers must choose between being summarized without enough traffic return or becoming less visible to readers entirely. A healthier market would let publishers negotiate those uses separately.
The ethical question is not whether summarization is useful. It often is. The question is whether the system can keep source economics alive. If AI answers consume reporting, analysis, reviews, and community discussion without sending enough value back, the supply of reliable source material weakens. That is not only a publisher problem. It becomes a search-quality problem, a civic-information problem, and eventually an AI-training problem.
The legal context is tightening. Publishers have sued AI companies over training data. Regulators are examining platform power. Cloudflare and other infrastructure companies are creating tools to identify and meter bots. Each of those moves reflects the same frustration: the web was built around human browsing and search referrals, while AI systems can consume far more content per user click. The economics no longer map cleanly.
Google still has a strong counterargument. Search has changed many times, from snippets to knowledge panels to voice answers, and publishers have often objected before adapting. Users like direct answers. If Google does not provide them, competitors will. That is commercially rational. But AI summaries are more comprehensive than earlier search features and can answer questions that previously required a visit to a source. The magnitude of substitution is different.
The next phase will likely involve deals, technical controls, and lawsuits rather than one dramatic break. Some publishers may demand usage-based licensing. Some may block AI crawlers. Some may make content more exclusive. Some may accept lower search traffic and build newsletters, subscriptions, events, or direct apps. The common thread is that dependence on Google as an audience lifeline now looks less stable.
The public should care because the outcome will shape what information AI systems can access and what information publishers can afford to produce. A search page that answers everything but weakens the reporting behind those answers is not a sustainable knowledge system. The companies negotiating this fight are arguing over revenue, but the result will decide how much of the open web remains worth publishing.
Topics: Google, AI Overviews, publishers, Reddit