Most online publishers can now pinpoint a time in the last eighteen months when the traffic graphs began to move in an unexpected way. There was content being released. There had been no collapse in search rankings. However, the clicks were not coming in as frequently as they once had. Without ever reaching the page, visitors were reading the headline in a search result box, finding what they were looking for, and leaving. The content had fulfilled its purpose. Nothing was received by the website.
This is how AI response engines have actually affected the online economy. The transaction that once supported the open web—user clicks on a link, publisher receives an ad impression, cycle continues—simply doesn’t occur when Google places an AI-generated summary at the top of a search results page or when Perplexity compiles a response from six sources and presents it as a single, clean answer. The data flows. The funds don’t.
It’s worth taking a moment to consider the magnitude of the change. Businesses such as independent publishers, local news sources, recipe websites, how-to manuals, and product review blogs are based on the idea that search engines provide traffic in return for content. For the better part of two decades, that agreement remained in effect. The fundamental premise of Google’s initial online proposal was that it would direct consumers to the top content. AI Overviews and similar products don’t make a big impression. The publisher is not there when they subtly revise the terms.
All of this is underpinned by the content scraping issue, which adds significant complexity. The content created by the same publishers that are currently experiencing a decline in traffic was used to train the AI systems that produced those on-page responses. The original outlet receives neither remuneration nor a visit when a news piece is written, indexed, scraped during training, and summarized in a search result. Anyone who has seen a big platform gradually extract value from a smaller environment will recognize this trend. The details are brand-new. It’s not the dynamic.
For the typical internet user, the response is predictable, albeit not especially encouraging. A growing number of publications are shifting important material behind paywalls. In recent years, The Financial Times, The New York Times, and other smaller publications have focused on increasing subscription revenue because they rightly predicted that search traffic-driven advertising would not be a stable source of income. That change is difficult but manageable for outlets with ample resources. The math is worse for smaller independent websites with narrower profit margins. Many are just shutting down, producing less, or doing both.
This feedback loop isn’t given enough attention. The quality of AI response engines depends on the content they were trained on and are now using. The quality of AI-generated responses will suffer if the economic conditions that drive content production collapse, resulting in fewer writers, fewer publications, and fewer people willing to devote time to creating something unique on the open web. The systems may find methods to make up for it, but it’s also feasible that the knowledge base they’re using begins to thin down in ways that manifest in the responses. Less publicly trainable data is produced by a paywalled, restricted web. Weaker responses result from less data. Users are eventually pushed elsewhere by weaker responses.
Slowly, regulators are starting to take notice of this. The EU’s AI and data use guidelines are beginning to raise concerns about whether unpaid content scraping qualifies as extraction that should be covered by the law. It’s a legitimate topic, and when the solutions are available, they will probably influence the future development of AI search tools. It’s a separate story completely whether it occurs quickly enough to affect the publishers’ current traffic loss.
The tempo of this play is actually unsettling to see. The open web is evolving more quickly than most people have realized. It is disorganized, ad-supported, and full of odd little websites that were made possible by search traffic. not completely vanishing. However, it was gently contracting in the direction of something more manageable and smaller than it was.
