Google pays Reddit roughly $60 million a year to license its content for AI training, according to Reuters. Then Google’s AI Overviews summarize that same content directly in search results — so users never need to visit Reddit at all. Publishers across the industry are finally doing the math on this arrangement, and the numbers look like a bad deal getting worse. The implicit bargain powering the open web is fracturing in real time.
The Traffic Bleed Is Real — and the Numbers Are Ugly
Independent research shows click-through rates plummeting as AI summaries answer queries before users ever reach the source.
Click-through rates drop 34% to 46% when AI Overviews appear, according to research cited by Search Engine Journal. Median Google referral traffic to premium publishers fell 10% year-over-year after the feature rolled out. That’s not a rounding error. That’s rent money — and it’s disappearing fast.
Reddit, USA Today, Reuters, Politico, The Economist, and People Inc. are all reconsidering how much access they give Google, per Wall Street Journal reporting summarized by Digiday. Reddit shares dropped as much as 5.8% in premarket trading after that story broke. Reddit’s own IPO disclosures revealed $203 million in aggregate data licensing contracts, underscoring how central selling content to AI companies has become to its business model.
By the numbers:
- 34–46% drop in click-through rates when AI summaries appear
- 10% year-over-year decline in median Google referral traffic to premium publishers
- $60M/year — estimated value of Reddit’s licensing deal with Google, per Reuters
- $203M — total aggregate value of Reddit’s data licensing arrangements, per IPO disclosures
“It’s time to take a stand and say enough is enough,” USA Today Co. CEO Mike Reed told Digiday. Think of it like a streaming service paying artists fractions of a cent per play while building a billion-dollar subscription business on their catalog. The math only works for one side.
The Catch-22 Nobody Wants to Say Out Loud
Google offers publishers controls for blocking AI training access — but using them risks losing broader search visibility entirely.
Google provides a tool called Google-Extended that lets publishers block their content from Gemini AI training. That sounds reasonable, except opting out of AI Overviews can still jeopardize broader search visibility — meaning publishers face a difficult trade-off between protecting their content and maintaining their search presence.
The UK’s Competition and Markets Authority recognized this conflict, ordering Google to let publishers opt out of AI search features without losing traditional search placement. Google said it would test those controls in Britain before a wider rollout, according to Social Media Today. That’s a notable regulatory precedent, even if the outcome beyond the UK remains uncertain.
Cloudflare CEO Matthew Prince has warned that AI-powered websites are gutting the web’s business model, with bots already accounting for a massive share of internet traffic. Google, for its part, counters that it still drives billions of visits to websites weekly and that click quality has improved. Publishers dispute both the magnitude and the interpretation.
What Comes Next
The content creators powering AI answers are reconsidering the deal — and what replaces it remains anyone’s guess.
What is unfolding here is bigger than a licensing spat. If the publishers and platforms powering AI answers start blocking crawler access or demanding compensation at scale, Google’s search products lose source material while publishers sacrifice visibility in exchange for control. The open web starts looking less like a public library and more like a delivery app that replaced the restaurant entirely. Europe already restricts Microsoft, Amazon, and Google from handling sensitive government data, signaling that regulatory appetite for reining in Big Tech is only growing. The old deal is already showing cracks — and nobody has agreed on what comes next.
