Google is quietly testing a pay-per-use AI licensing model for publishers, but early participants describe the pilot as a “black box” that is more promising as a precedent than as a revenue line.
The program, an “AI contribution pilot” connected to Google Search Console, pays publishers when their content contributes to AI-generated responses across Gemini, AI Overviews and AI Mode. Once a publisher opts in, they see an AI earnings widget showing a monthly figure with some history — yet no detailed breakdown of how the payout is calculated. Payments are usage-based, carry no upfront fee, and publishers can opt out at any time.
What’s actually being tested
Google confirmed the scheme is an early-stage learning pilot to test how best to reward high-quality content. The company pointed back to a June blog post about working with websites whose content helps keep generative AI answers fresh and accurate. According to Digiday, at least dozens of publishers have been approached, with the program proving more attractive to small and mid-sized publishers than large ones.
For broader context, Google says its news AI scheme already involves more than 200 titles globally, while its Google News Showcase licensing programme covers over 2,800 publications in 33 countries.
Search Console screenshots seen by Digiday show an “AI contribution” panel with a monthly payout figure, but little information about the calculation. “It’s quite black box,” one executive with knowledge of the matter told Digiday.
Still, two publisher executives in the pilot said basic data is better than sitting on the sidelines. One noted that weekly calls with Google have been “extremely collaborative” and sees the effort less as a one-off deal than an early test of a marketplace for inference data.
Why the industry is divided
The pilot lands at a delicate moment. Google launched an AI performance report within Search Console in August, and one publishing executive said the new AI-surfaces reports closely track traffic lost from traditional search and Discover. That raises the question: if AI exposure replaces search traffic, how should that exposure be compensated?
Smaller publishers may find it hard to refuse recurring revenue for content they are already producing. But early returns have been called “peanuts” relative to advertising revenue, and one publishing exec described the offers as “lowball numbers.” The bigger concern is strategic: participation could weaken publishers’ leverage to demand better terms later, because Google can point to the program as compensation.
Luke Stillman, managing director of Madison and Wall, told Digiday that publishers have relatively little leverage over how AI changes discovery and distribution. His advice: create a new revenue stream where possible and use it to fund businesses less dependent on advertising, such as subscriptions and events.
What publishers should watch
- If invited, treat the pilot as a low-risk way to see AI contribution data — not as a core revenue line.
- Track AI performance reports alongside traditional search and Discover declines to build a clearer picture of substitution.
- Use any initial payout to diversify into subscriptions, events or first-party data products rather than relying on referral economics.
- Document how often AI answers reference your content to strengthen future licensing conversations.
David Buttle, founder of Spur and DJB Strategies, argues that per-use payments look less like a meaningful exchange and more like a strategic hedge. Google “doesn’t want” a market where it pays based on actual journalism usage in AI, he said, since that would be “the thin end of the wedge” for its search model. But the pilot at least introduces the idea of usage events that convey value — and a return path of value back to publishers.
For marketing and publishing teams, the message is to stay close to this experiment. The economics may be small today, but the program could signal how Google plans to compensate content in an AI-first search environment, which will shape content budgets, SEO and ad-supported publishing models over the next few years.
Source: Digiday
