Publishers say AI data brokers are becoming ad tech’s next costly middle layer

Publishers have spent the better part of the past decade arguing over one stubborn problem: too many intermediaries take too much out of digital advertising.

Now, as AI companies race to gather, structure and monetize content and audience data, many media executives see a familiar pattern coming back in a new form.

This time, the concern is not just the classic ad tech tax. It is the rise of AI data brokers and other intermediaries that sit between publishers and the companies building products on top of publisher-generated information. The fear is straightforward: these middlemen may be able to keep most, or even all, of the upside.

For publishers, that cuts deeper than a revenue-share dispute. It raises questions about who controls access to their content, who profits from audience intelligence, and whether media companies are once again supplying the raw material while someone else builds the more lucrative business around it.

The complaint lands at a moment when publishers are already trying to defend margins across ads, subscriptions, licensing and platform distribution. AI adds another layer of urgency because the market is still being formed in real time.

That creates an opening for intermediaries. Companies that aggregate content, normalize data, broker licensing relationships or package publisher information for AI use can become the connective tissue between media owners and model builders. In theory, that can reduce friction. In practice, publishers worry it can also reduce their leverage.

Part of the anxiety comes from history. In programmatic advertising, publishers watched a long chain of vendors emerge between advertiser demand and publisher inventory. Each link in that chain promised efficiency, targeting or scale. The result, publishers have long argued, was a fragmented market where too much value leaked out before revenue reached the content creator.

Many now see AI repeating the same structural logic. If outside companies are best positioned to collect, process and sell access to publisher data, then the publisher risks becoming the least powerful party in a market built on its own assets.

That does not mean all intermediaries are unnecessary. Some publishers will likely need partners to manage rights, structure archives, validate permissions, handle metadata or package content for AI systems. The issue is less the existence of middlemen than the terms on which they operate.

Publishers want clearer answers on what is being licensed, how data is being used, whether attribution is preserved and what kind of compensation flows back. They are also increasingly alert to the possibility that audience signals, contextual data and proprietary content relationships could become long-term strategic assets in the AI economy.

If those assets are bundled and resold by third parties, publishers may lose more than near-term revenue. They may also lose direct visibility into how their material creates value elsewhere.

That is why the current debate is broader than scraping or one-off licensing deals. It is about market structure. Who gets direct access? Who sits in the middle? Who owns the commercial relationship? And who ends up with pricing power once AI supply chains mature?

For larger publishers, the answer may be to negotiate direct deals wherever possible and to be more selective about external data partnerships. For smaller publishers, the tradeoff is harder. Intermediaries can offer scale and convenience, but they can also consolidate control in ways that become difficult to unwind later.

The next phase of this fight will likely revolve around transparency and ownership. Publishers are not just looking for checks. They want enforceable terms around training access, downstream usage, attribution and data governance. Without that, AI partnerships risk becoming another ecosystem where publishers provide the inputs but do not shape the economics.

There is also a competitive angle. If a small number of brokers become the default gatekeepers for publisher data, they could gain outsized influence over what gets surfaced, licensed and valued. That would give publishers a problem they know well: dependence on infrastructure they do not control.

The difference now is speed. AI markets are moving faster than many previous platform shifts. That leaves less time for publishers to establish standards before commercial habits harden.

In other words, this is not just another debate about vendor fees. It is an early warning about who captures value in the next version of the media supply chain.

Publishers have seen this movie before. The only open question is whether they can change the ending before the new middle layer becomes permanent.