Google Is Paying for AI Answers. Almost Nobody Gets Paid.

Google started paying publishers for content used in its AI answers.

Read that sentence again, because it is real. Google confirmed an early-stage pilot to Digiday on September 14 that pays selected publishers when their content "contributes significantly" to answers generated in AI Overviews, AI Mode, or the Gemini app. Participants see an "AI earnings" panel inside Search Console. Google's own description of it reads: "Accumulate earnings when your content contributes significantly to AI-generated responses in participating products."

The headlines made this sound like a turning point. Read the fine print and it reads like a pressure valve.

What the pilot actually is

Roughly 100 digital publishers are enrolled, according to reporting from The Information. That is it. Out of the entire open web, about one hundred sites get a monthly line item.

And the payouts are all over the place. One publisher that joined near the start has earned more than $1 million over a year. Another received roughly $50,000 to $60,000 over a few months. Meanwhile several small and midsize sites have collected less than $1,000 across several months, which amounts to less than 0.1 percent of their ad revenue.

The formula is a black box. Participants told The Information they do not know how Google calculates the payments, and amounts can change month to month without explanation. The Search Console widget shows a monthly payout figure. It does not show which pages earned it, how often your content was used, or why the number moved.

One pattern did surface: content on niche topics with strong followings, anime and gaming among them, appears to earn more than average.

The catch everyone skips

Here is the part that matters most, and it is buried in the opt-out wording.

You can opt out of the payment. That does not opt you out of the usage. Google's own documentation says that if you leave, content from other sites will still appear in those AI features and, in Google's words, "may appear similar to yours." Your question does not disappear. Your competitor just answers it instead.

So the deal on the table is this: take whatever number Google assigns you, or take nothing, and Google uses the content either way.

Some larger publishers are refusing to join the program to pressure Google into paying more, according to The Information. Their traffic is already falling, and the legal pressure is real. Penske Media, the parent of Rolling Stone, sued Google in September 2025 over lost traffic and ad revenue. The European Commission opened an antitrust investigation in December 2025 into how Google uses publisher content for AI features without adequate payment or a genuine way to opt out. A German court has ruled that AI Overviews are Google's own content, not summaries of existing material, which gives publishers a basis to demand licensing fees for AI answers that draw on their work.

That is why this pilot exists. Not generosity. Leverage management. The Decoder described Google's approach as divide-and-conquer: individual deals keep publishers negotiating separately instead of collectively, a few get paid, most get little or nothing, and the ones that walk away change nothing because other sources fill the gap.

What I would do instead of waiting for a check

I build AI visibility campaigns for a living. One of our GEO campaigns at First Rank moved LLM search traffic up 98 percent in 30 days for a client, and not a single dollar of it came from a Google payout. It came from citations. Being the source the AI names, links, or quietly draws on is the part that compounds, because the citation drives the visit, and the visit is something you can measure and monetize yourself.

Google just made the measuring part easier, which is the actually useful news this week.

Gemini now adds UTM parameters to its outgoing links. Google's John Mueller confirmed the change on Reddit and asked the community for more detail on how referrers behave. Here is why it matters: Gemini traffic used to collapse into your "direct" bucket in GA4 because referrers get stripped in mobile apps and in-app webviews. A UTM survives that trip and shows up in your raw server logs too.

So do the unglamorous work:

1. Check your GA4 AI Assistant channel, the default channel group Google added in May 2026 that covers ChatGPT, Gemini, Claude, and others. Treat that number as a floor, not a total.
2. Segment AI referral traffic by landing page, not just session counts. A referral to your service page and a referral to a blog post are not the same number.
3. Remember that AI Overviews and AI Mode clicks land in Organic Search in GA4, not in the AI channel. No single channel covers everything Google's AI features send you.

And keep investing in the thing that earns citations in the first place: pages that answer questions directly, specifics instead of adjectives, tables a model can lift a fact from. I wrote the full playbook on that a few days ago.

The scale argument still favors doing this yourself. AI referrals were roughly 1 percent of visits across nearly 14,000 domains measured by Conductor, with the vast majority coming from ChatGPT. The Decoder points to a Pew study finding that only 1 percent of users click a source link in AI Overviews. The checks Google is writing will not replace that traffic for you. Being cited might.

The bottom line

Google priced your content. The price is pocket change for small sites, the formula is secret, the payouts wobble without explanation, and leaving the program changes nothing about how your content gets used.

Do not build your strategy around the payout. Build it around the citation. The payment that scales is the click.