Google's AI search is eating publisher traffic alive

For decades, Google had a simple deal with the web: it indexed your articles, and in exchange, it sent users to your site via ten blue links. That bargain is now officially crumbling.

According to a report from The Wall Street Journal, Google’s AI Overviews and conversational chatbots are dealing a devastating blow to referral traffic for news publishers.

When searchers get an AI-generated summary right at the top of the results page, often synthesized straight from newsrooms without attribution or consent, they have zero reason to click through. The traffic that supports original reporting is simply evaporating.

The blue links are disappearing fast

Google introduced AI Overviews last year. The feature initially devastated niche sites focused on travel itineraries, wellness advice, and product reviews.

Now, mainstream news organizations are feeling the pain. Data from Similarweb cited by the Journal shows that organic search accounted for just 36.5% of desktop and mobile visits to The New York Times in April 2025, down sharply from 44% three years earlier.

Things will likely get worse before they get better. Google's conversational AI Mode, designed to take on ChatGPT, relies on a fluid chat interface that offers even fewer outbound web links.

Google, naturally, paints a rosier picture. At its developer conference in May, the company claimed AI Overviews actually increased search traffic. But for publishers on the receiving end of the referral drought, that traffic stays firmly inside Google's walled garden.

Licensing deals become a survival tactic

Publishers know the status quo is dead. Executives at The Atlantic and The Washington Post have openly stated that the media business must pivot its revenue models immediately.

To cushion the blow, several publishers are rushing to cut data-licensing deals with big tech firms. The New York Times recently inked an agreement with Amazon to license its editorial archives for AI model training.

Meanwhile, outlets like The Atlantic have partnered with OpenAI, and AI search startup Perplexity is trying an ad-revenue-sharing model that pays publishers whenever its bot references their stories in answers.

Why it matters

The fundamental economics of the open web are being rewritten in real time. For over two decades, search engines acted as digital switchboards connecting readers to creators; today, Google is transforming into an answer engine that hoards the audience for itself.

In our view, licensing news content to giants like Amazon and OpenAI may provide quick cash, but it looks like a temporary bandage rather than a long-term cure. Once a model is trained on quality journalism, publishers lose their leverage, and a one-time licensing check cannot replace a steady stream of loyal direct subscribers.

Perplexity's attempt to share query-level ad revenue is clever and feels fairer to creators, though it remains unproven at scale. Ultimately, publishers can no longer count on SEO to keep the lights on. Survival now means forcing readers to care about the brand enough to visit directly, or accepting a future as invisible training fodder for Silicon Valley.