The emerging market for publisher content is beginning to take shape. Large media companies are negotiating licensing agreements with AI platforms, building internal products and establishing terms for how their journalism can be used. That progress risks obscuring a harder problem.
Most publishers will never negotiate these agreements on their own.
Recent research has found that AI licensing remains concentrated in a relatively small number of individual contracts. Industry insiders have worried that selective dealmaking could reinforce existing structural inequalities, leaving smaller outlets overlooked because technology companies may not recognize the value they provide to particular communities.
The open web is not made up only of national newspapers, global broadcasters and major magazine groups. It depends on local newsrooms, trade publications, independent magazines, nonprofit outlets, specialist reviewers and community publishers. These organizations often produce the most specific and useful information available on a subject. They know the local culture or the technical category in a way that a general-interest publisher cannot.
Why Small Publishers Matter to AI
This is precisely the kind of material AI systems need. The usefulness of an AI product depends heavily on the quality and breadth of the information it can access. General knowledge can answer common questions. More demanding tasks require sources with deep expertise, current reporting, and context that cannot be reconstructed from large publishers alone.
Yet the organizations producing that information are also among the least equipped to participate in the developing AI economy. They may lack dedicated licensing teams, legal resources, or technical infrastructure. Even when an AI company is willing to pay for access, the cost of identifying, evaluating and contracting with thousands of smaller publishers can exceed the value of any individual agreement. That creates a market failure.
A Market That Doesn’t Connect
AI companies need access to a wide range of trusted sources, while smaller publishers need a practical way to license their work. The current deal-by-deal model does not connect them efficiently. If this problem remains unresolved, the consequences will extend beyond publisher revenue. Value will increasingly concentrate around the largest content owners because they are the easiest to find and contract with.
Smaller publishers will see their work used indirectly, summarized without meaningful compensation, or excluded from licensed AI products altogether. None of those outcomes is sustainable. A licensing market dominated by a few major publishers would also make AI less useful. It would narrow the range of voices and expertise available to users.
A Thinner Knowledge Base
Local events would be interpreted through national coverage. Specialized industries would be represented by broad summaries rather than practitioners and reporters who understand them. Communities that already receive limited media attention would become even less visible. Over time, the problem compounds.
Other recent research has shown that publishers receiving between 1,000 and 10,000 daily page views lost 60 percent of their traditional search referrals over two years, compared with 22 percent for large publishers. Meanwhile, Cloudflare has documented a widening imbalance between the amount of publisher content AI services crawl and the traffic they return. Its 2025 data showed some AI platforms retrieving hundreds or thousands of pages for every visitor referred back. The old assumption – that exposure will repay publishers for access – is becoming untenable.
When smaller publishers cannot earn sufficient returns from producing original work, they publish less of it. AI systems then have less credible information to draw from. The technology may become more capable while the knowledge base beneath it becomes thinner. The industry needs a model that allows smaller publishers to participate without requiring each of them to become a sophisticated licensing operation.
What a Fair System Requires
That begins with shared standards. Publishers need a consistent way to define what content is available, under what conditions, and for which uses. AI systems need a reliable way to recognize those permissions and distinguish licensed material from restricted content. Publishers also need transparent reporting. They should be able to see when their work is accessed, how it contributes to an AI experience, and how compensation is calculated.
Without that visibility, licensing risks becoming another opaque digital marketplace in which smaller participants have little ability to verify value. Payment mechanisms must also reflect the realities of the long tail. A small trade publication may never generate the volume of a major newspaper, but its reporting may be uniquely valuable for a high-intent query. Compensation should recognize relevance and utility along with scale.
Most importantly, participation must be technically simple. The average publisher should not need to maintain custom integrations or negotiate separate rules with every AI company. Rights management has to operate as infrastructure, not as a bespoke professional service.
The Exchange at Stake
The open web has always depended on an economic exchange. Publishers invested in reporting and expertise because audiences, advertisers and subscribers created value in return. AI is changing how that information is discovered and consumed, but it does not eliminate the need for that exchange. The question is whether the nascent system will support the full range of publishers that make the web useful or only the institutions large enough to negotiate their own terms.
The largest publishers will find ways to participate. The future of the information ecosystem may depend on whether everyone else can, too.
