Right now there are only two ways publishers get paid for AI content: be large enough that a platform offers you a contract, or have nothing and threaten to leave. On October 30 Meta's new commercial terms make crawling your website a condition of keeping a Facebook or Instagram business account. The same week, Google confirmed it is paying more than 400 publications, Forbes among them, for expanded rights. If you run a publisher that is not on that list, this piece is about what the gap means and how it closes.

  • From October 30, 2026, keeping a Meta business account counts as permission for Meta to crawl the website linked to it; the only way to say no is to delete the account.
  • Google's paid News AI pilot started in December 2025 and now covers more than 400 publications; you cannot apply, you get invited, and the rates are secret.
  • On the old web every publisher got a price for its pages through an ad auction; on the AI web the big get a contract and everyone else gets a terms update.
  • The way a small publisher gets paid for AI content is the same way it got paid before: brands bidding for its pages, this time the pages AI engines cite.

How do publishers get paid when AI uses their content?

Getting paid for AI content means receiving money when an AI engine such as ChatGPT, Google AI Overviews or Perplexity reads your pages and uses them to build an answer, whether or not anyone clicks through to your site. In October 2026 that money arrives in two forms only: a licensing contract from a platform, offered to publishers with enough scale, or nothing. A third form, brands paying to appear on the pages AI cites, exists but is not yet organised, and it is the one a small publisher can actually reach.

How publishers got paid on Google before AI, and what changed

Before AI, size did not decide whether you got paid. A site with 50,000 monthly readers could not negotiate with Google, but AdSense put a price on its pages every day because advertisers bid for the readers on them. Special programmes existed for the large (AMP, Instant Articles, News Showcase), but the long tail of publishers was paid by advertiser demand, not by invitation.

AI search flipped that. The 400 publications in Google's pilot get a contract; everyone with a Facebook Page gets a terms update. There is no auction between the platform and the small publisher any more, so your influence on an AI answer is worth whatever the single buyer decides, which for most publishers is zero.

What Mediabistro reported about Meta and Google, and what we checked

Mediabistro's October 20, 2026 roundup describes the week as a split: Meta claims AI training rights over content posted after October 30, Google pays Forbes for access. Two corrections from the original sources. Press Gazette's October 19 report prints the actual clause: it covers the domains and URLs linked to a business account, meaning your website, not just your posts. And Meta has since removed the word 'train' while keeping 'develop, improve, research', which US media lawyer Matt Topic told Nieman Lab is broad enough to include training anyway.

On the Google side, Digiday's Jessica Davies reported on October 20 that Forbes chief innovation officer Nina Gould confirmed the deal is paid but would not say how much, that Forbes' referral traffic is down about 45% year on year, and that the programme launched in December 2025 with the Guardian and the Washington Post before passing 400 publications this month.

Is it better to sign a Google AI deal or just leave Meta?

Why Forbes signed with Google, and why Meta says you can stay

Gould's reasoning is simple: Forbes gets paid, gets data on how its articles perform inside Google's AI features, and skips the staff and computing cost of testing that itself. Reach CEO Piers North told Digiday on October 16 that his time-limited Google deal is a template for other AI agreements. Meta's statement to Press Gazette argues the other side: the terms only cover ads and business products, robots.txt still signals your preferences, and nothing about its crawling has changed. On that reading, you take the contracts you can get and keep the reach you cannot afford to lose.

Why neither option pays a small publisher

Both halves of that advice assume you are big. Forbes was invited; the advice for everyone else, from the News Media Association via Press Gazette, is to delete every Meta business account before October 30, while the Movement for an Open Web prepares a complaint to the European Commission and the UK CMA. Leaving is real leverage, but you can only use it once, and it earns you silence, not revenue.

A contract is better, but it is still a number set by the only buyer, shown to no one and renewed when the buyer feels like it. Forbes is already doing the thing that puts a real price on its content: Digiday reports it uses citation data to sell affiliate, sponsorship and native deals to brands that want to show up in AI search. That is advertiser money finding the cited page by hand. The AI web needs that done automatically, for the publishers Google will never call.

What this means if you run a publisher, and what it means for brands

For publishers: what to do before October 30, and what not to expect from it

Check your robots.txt and list every Meta business account before October 30, and treat the clause as a legal exposure, not an annoyance. Then stop waiting for an invitation. Find out which of your pages AI engines cite for queries where people are about to buy something, because those pages are your sellable inventory. A publisher that can show brands demand for its citations has leverage with every platform, whatever its size.

For brands and agencies: the pages Google pays for are the pages to buy

The publications in Google's pilot are, by definition, the ones shaping AI answers in your category, and brands sit in none of those deals. Map which publishers AI engines cite for your commercial queries, then buy presence on those pages the way Forbes' brand partners already do. Keep that budget separate from paid search: it buys the source of the recommendation, not the ad slot underneath it.

What happens after October 30: three things to watch

First, whether the European Publishers Council's letter to Brussels and the Movement for an Open Web complaint push Meta to reissue the terms with an opt-out; a forced opt-out would become the baseline for every platform. Second, whether Google's 400-publication pilot ever publishes a rate; the first public number becomes the floor every other buyer prices against. Third, how many publishers copy Forbes and sell citation data to brands directly; once that is routine, the advertiser side has arrived and the question becomes who standardises it.

Conclusion

Hold on to this: how publishers get paid for AI content today depends on size, not on how useful their content is to the answer, because nothing yet puts a market price on the citation itself. Smalk AI's answer is Generative Engine Advertising, native ads for AI agents placed on the publisher pages AI engines cite, so brands buy presence where recommendations form and any publisher, invited or not, earns from advertiser demand rather than from a terms update. What to watch next: the first public price for a citation, whether it comes from Brussels, from Google's pilot, or from a brand-funded market.