Abstract:
Google is paying about 100 digital publishers to compensate their content for AI-generated answers in search and other products, according to people familiar with the testing. The pilot program marks a further shift by Google toward paying publishers to display their content in its mainstream search engine.
In the past, Google's position was that it should not pay for content from publishers that answered user questions, on the grounds that searches would drive traffic to these publishers' websites. However, after the advent of AI Overviews, Google search traffic directed to external websites dropped sharply, causing widespread dissatisfaction among publishers, and some publishers even filed lawsuits over this.
However, there is a high probability that this plan will not quell the above dissatisfaction. Several people familiar with the payment situation said that for many small and medium-sized blogs and websites participating in the pilot, this payment is not even one thousandth of their advertising revenue. Three people familiar with the matter said the meager scale of the revenue has discouraged some large publishers from participating in the project. These large publishers are choosing to wait and see, hoping to pressure Google and others to pay more for using AI content.
Some publishers also see this pilot as an opportunity to find out what content Google AI values and then adjust their creative direction accordingly. A publisher executive involved in the project said that Google seems to be trying out a new cooperation model: taking the publisher's content, and then Google unilaterally determines the value of this content.
A person close to Google revealed that Google launched this paid pilot less than a year ago for AI summaries, AI patterns, and Gemini chatbots. There is a huge disparity in the compensation received by each publisher: a publisher that joined the pilot a few months ago has received a total of about US$50,000-60,000, which is a small proportion of its business; another publisher that joined the pilot earlier can receive more than US$1 million per year, which accounts for a large proportion of its revenue; and some small websites have a total income of less than US$1,000 in a few months.
Publishers can track the frequency of their content being called and the compensation they receive through Google Search Console. This tool is used to count and monitor website performance and search traffic.
If enough publishers continue to participate, this model may become a reference model for other AI companies to deal with web content - the content of these websites is the source of the answers generated by the AI model.
Industry media Digiday first exposed the existence of the pilot, but did not disclose details such as the payment amount and the number of participating publishers.
Waves of protests from publishers
Google’s attitude toward paid collaboration has been evolving. After facing pressure from regulators and lawmakers in many countries, Google has signed licensing and cooperation agreements in recent years to pay for the use of publishers' content in its own products. After technology companies started AI model training, the situation changed again. For a while, Google was reluctant to pay for the use of publisher content for AI training, but in 2024, Google signed a deal to obtain Reddit data for use in training the Gemini large model.
In January 2025, Google reached a cooperation with the Associated Press to connect the Gemini application to the Associated Press real-time news information flow. In recent months, Google has also negotiated with publishers such as The Washington Post and The Guardian to pay to give these media content more exposure in the AI article summarization module of Google News.
The background to the introduction of this AI summary paid plan is publishers’ continued complaints: Google’s AI products directly summarize web page information, resulting in users no longer needing to jump to visit publisher websites. The Encyclopedia of World History, an online encyclopedia website that did not participate in this pilot, stated that in the past two years, the website’s traffic from Google has declined by 25%-35% each year, compared with the overall traffic drop of approximately 50% in 2024; founder Jan van der Klaben said that the content section where AI summaries appear on the page has been hardest hit.
Like other large model manufacturers, Google cannot do without online publishers to continue to produce content. The timeliness of large models trained based on Internet data will continue to decrease. In order to ensure the timeliness and accuracy of answers, the model needs to retrieve the web content of Google search for verification and correction, so as to reduce "illusion" - that is, the phenomenon of large models seriously outputting completely wrong answers.
In this pilot, not all page sources are weighted equally. For example: the model retrieves 100 websites to generate an answer, and the content of only 5 of them (5%) may actually improve the quality of the answer.
Some publishers participating in the pilot said that they are still unclear about the calculation logic of remuneration, what counts as "effective content contribution" and what type of content has the highest value; and remuneration will fluctuate month by month without warning.
But some patterns have emerged: People familiar with the matter revealed that some topics that are less reported on the Internet but have strong user demand, such as animation and game-related content, can receive higher remuneration.
As Google presses ahead with this experiment, regulators are stepping up pressure to scrutinize the way Google uses publisher content. Penske Media launched two lawsuits against Google within four months: one accused the AI summary function of diverting website traffic, cutting affiliate marketing revenue, and illegal behavior of forcibly binding index inclusion and AI summary display; the other relied on the court's ruling that Google had an illegal monopoly in the field of advertising technology, and demanded compensation for related losses. Google's policy at the time was that publishers could restrict some specific uses of their content, but if they refused AI calls, they would lose exposure in Google searches.
Google denies all accusations. Its defense reasons include: publishers can completely prevent Google from crawling and indexing their own websites; advertisers and publishers also have other advertising products to choose from, and they choose Google only because it is more cost-effective.
In June this year, British regulators asked Google that when publishers opt out of AI search features, their ranking in ordinary search results should not be affected. The European Commission has also launched an antitrust investigation in December 2025 to verify whether Google has unfair terms - using publisher content for AI functions without giving publishers reasonable compensation and without a real right to refuse.
Following regulatory pressure from the UK and the EU, Google officially launched a publisher opt-out option in August: publishers can refuse to use their content for AI summaries, AI models, and Google discovery information flows without losing search exposure.
Comments