Google has struck content licensing deals with roughly 100 digital publishers to train and power its AI products, but the payments reveal a stark disparity between how the company values different sources. According to The Information, compensation ranges wildly from less than $1,000 over several months for smaller outlets to more than $1 million annually for major publishers. The median payment sits far closer to the bottom of that spectrum.
The deals cover content used across Google's AI Overviews (direct answers embedded in search results), AI Mode, and Gemini chatbot. Google frames these agreements as voluntary partnerships, yet the payment structure exposes how little leverage most publishers hold. Many smaller outlets accepted whatever Google offered without understanding the calculation methodology. Some larger publishers have rejected initial offers and are demanding higher rates, but Google's ability to index their content without explicit permission creates pressure to accept unfavorable terms.
This payment model reflects a broader power imbalance in the AI economy. Google generates billions in advertising revenue from search, and AI Overviews directly present answers on Google's platform rather than directing users to publisher websites. When a user gets an AI-generated summary instead of clicking through to a news site or blog, the publisher loses traffic and potential ad revenue. Google's licensing payments are meant to offset this loss, but the amounts suggest the company values content creation far below its actual contribution to its AI systems.
The secrecy around payment calculations compounds the problem. Publishers cannot benchmark against peers or negotiate from a position of knowledge. Google controls both the valuation and the terms, while individual outlets have minimal bargaining power. Larger publishers with substantial archives and traffic have more leverage, but even they face a choice between accepting Google's offer or risking exclusion from AI products that increasingly drive discovery.
The approach differs sharply from how Google historically treated publishers. The search giant famously fought news publishers over snippet licensing and aggregation rights, ultimately leading to regulations like Australia's News Media Bargaining Code and the EU's Copyright Directive. Those fights forced Google to negotiate payments for news content in certain regions. Yet with AI, the company appears to be pre-empting similar battles by offering deals early, albeit on highly favorable terms.
The stakes extend beyond immediate revenue. As AI becomes a primary discovery mechanism rather than search, publishers that accepted low-ball deals early lock in disadvantageous terms while Google's AI products grow more valuable. Publishers excluded from deals lose visibility in AI systems altogether. The result is a fragmented market where Google controls both the technology and the negotiating framework.
Some publishers view the licensing deals as necessary insurance against potential litigation over copyright infringement in AI training. Others see them as insufficient compensation for content that drives engagement and accuracy in Google's products. Smaller outlets struggle most, lacking both the leverage to demand better terms and the legal resources to challenge Google's practices.
This model will likely face regulatory scrutiny. Policymakers in Europe and elsewhere have begun examining how AI companies acquire and compensate for training data. Google's current approach suggests the company believes modest payments and selective licensing will satisfy future regulatory requirements. Whether that calculation holds depends on whether publishers organize collectively and whether regulators view current arrangements as exploitative.