Small news publishers who refuse AI licensing deals lose search visibility, while those who sign get pennies — creating a coercive two-tier internet

legal0 views
By early 2026, the AI licensing landscape for news publishers has split into haves and have-nots. Large publishers like News Corp ($50 million/year from Meta), the New York Times (deal with Amazon), and the Associated Press (deal with Google) have signed multi-year licensing agreements. But small and mid-size publishers face an impossible choice: sign a licensing deal for a fraction of their content's value, or refuse and watch their search traffic evaporate as AI-generated summaries replace click-throughs. Google's AI Overviews present the most acute version of this problem. When Google uses a publisher's content to generate an AI Overview that answers a user's question directly on the search results page, the user never clicks through to the publisher's site. Publishers who opted out of Google's AI training crawlers discovered that Google still uses their content for AI Overviews through its regular search index — there is no way to opt out of AI Overviews without opting out of Google Search entirely. For small publishers who depend on search traffic for 40-70% of their audience, this is not a choice at all. The Digiday publisher scorecard showed that most publishers rated Big Tech's licensing deals poorly, viewing them as inadequate compensation for the value being extracted. This problem persists because of an extreme power asymmetry. There are a handful of AI companies (Google, OpenAI, Meta, Amazon) and thousands of publishers. Each individual publisher's content is dispensable to an AI model trained on billions of pages, but the AI platform's search traffic is indispensable to the publisher. This gives AI companies overwhelming negotiating leverage. The RSL Collective and IAB Tech Lab's CoMP framework are attempting to standardize licensing terms, but standardization only helps if publishers have the collective bargaining power to enforce those terms. The New York Times and Chicago Tribune sued Perplexity in December 2025 for copyright infringement, but most small publishers cannot afford litigation. The result is a two-tier internet: large publishers get paid (poorly), small publishers get scraped for free, and the journalism that communities depend on becomes economically unviable.

Evidence

News Corp deal with Meta at $50M/year (https://digiday.com/media/meta-enters-ai-licensing-fray-striking-deals-with-people-inc-usa-today-co-and-more/). Google uses opted-out publisher content for AI Overviews (https://www.niemanlab.org/2025/05/google-is-using-content-from-publishers-who-opt-out-of-other-ai-training-to-power-ai-overviews/). Timeline of publisher-AI deals in 2025 (https://digiday.com/media/a-timeline-of-the-major-deals-between-publishers-and-ai-tech-companies-in-2025/). NYT and Chicago Tribune sued Perplexity (https://digiday.com/media/here-are-the-biggest-moments-in-ai-for-publishers-in-2025/). RSL Collective and CoMP framework (https://digiday.com/media/publishers-scorecard-for-big-techs-ai-licensing-deals/).

Comments