There’s a pattern publishers have learned to recognize. A platform sends you traffic, you build a business on it, the platform pulls the traffic back, and then it offers to sell some of it back to you on new terms. This week The Information reported the latest turn of that wheel: Google is taking a tough line with news publishers over AI content licensing, and it’s using money publishers already receive as the lever.

Before the cynicism takes over, it’s worth being precise about what is confirmed and what is one outlet’s reporting. Those are not the same thing, and the difference matters if you’re a publisher deciding how to react.

What Google is offering

In December, Google launched a News AI pilot program with a small group of partners, including The Washington Post, The Guardian, and Der Spiegel. The pitch: AI-generated article overviews and audio briefings inside Google News and Gemini, each carrying source attribution and links back to the original article. Google frames it as a way to “drive engaged audiences back to original news outlets.” That framing is Google’s own, and it’s the part the company is happy to say out loud.

On its face, that’s not a bad offer. Visibility inside AI surfaces, with a credited link, is exactly the kind of placement publishers worry they’re losing as AI answers absorb the clicks that search used to send.

The part that’s reported, not confirmed

Here’s where the asterisk goes. According to The Information, the price of entry is broad rights to use editorial content for AI purposes, including the ability to train Google’s AI models. That detail is the heart of the story, and it’s also the part to handle carefully: it comes from the reporting and a person who worked on the project, not from a Google announcement. Google did not state a model-training requirement explicitly. The connection journalists draw leans on Google’s existing language about “commercial partnerships” and “publicly available” data.

So the honest version is: a credible outlet reports that broad AI-training rights are on the table, Google has not confirmed that framing, and a flat fee is reportedly what’s on offer in exchange. Some publishers are hesitant precisely because extensive rights for a flat rate is a poor trade when nobody yet knows what those rights will be worth.

The leverage, and Google’s pushback

The reason this lands harder than a normal licensing pitch is the leverage. The Information reports that Google has told some publishers it intends to let Google News Showcase expire. Showcase is the older program through which Google pays publishers for curated content. If it winds down, the reporting goes, publishers who decline the new AI program would eventually lose those payments too. Join on Google’s terms, or watch the existing check disappear.

Google disputes this. A spokesperson said Showcase agreements continue to be extended, contradicting the reports of a planned shutdown, and the company’s public line is simply that it is “expanding our partnerships through our News AI pilot program.” Both things are now on the record: the reported pressure tactic, and Google’s denial that Showcase is ending. A reader is entitled to weigh them, but pretending the wind-down is settled fact would be dishonest. It isn’t, yet.

Why this is more than a big-publisher story

It’s easy to file this under enterprise drama. The names in the story are national newsrooms with lawyers and licensing teams. Most of the publishers we work with are smaller: regional sites, niche trade outlets, diaspora media serving readers across the EU. None of them are getting a Showcase check or a seat in the December pilot.

But the structural lesson is the same at every size. The traffic publishers built on is moving inside AI answers, and the terms of access there are being written right now, by the platform, in private negotiations. Publishers say search referrals have fallen sharply since AI Overviews arrived, some describing roughly half their search traffic gone. When your distribution and your revenue both run through one company, that company sets the price, and it can reset the price whenever the product changes. That isn’t malice. It’s leverage, and leverage compounds.

The small-publisher version of protecting yourself isn’t a licensing negotiation you’ll never be invited to. It’s three quieter decisions. Own the audience relationship you actually control, through email, subscriptions, and membership, so a single algorithm change can’t halve your business overnight. Decide deliberately what AI systems may do with your content, at the level you can enforce, rather than discovering your default settings made the choice for you. And under GDPR, treat any audience-data or subscription mechanic as a consent question first, because for EU and diaspora outlets the compliance side is the entry ticket, not the polish.

What to actually do

If you publish at any scale, this is a prompt to read your own posture honestly. Know what your robots and AI-access controls currently allow, because that is the only “licensing negotiation” most publishers will ever have, and many are agreeing to things by inaction. Build the direct channels that don’t depend on Google’s distribution, the same logic behind surviving as a publisher in the AI-first era and behind Google’s own Subscription Linking push, which rewards a verified relationship with your readers. And if you want to be cited and credited inside AI answers instead of merely scraped into them, the work of being visible and quotable in AI search is the same work, whether or not Google ever offers you a contract.

The publishers in the headlines are negotiating over a flat fee. The rest of us are negotiating, by default, over our terms of service settings. Both are worth paying attention to this week.

Sources