AI royalties are lastly on the desk for small and midsize publishers.
Don’t credit score OpenAI’s ChatGPT, credit score company LLMs — enterprise RAG is what’s creating royalty income for publishers.
RAG — retrieval augmented technology — kicks in when a consumer (or system) prompts the LLM, which then pulls the related content material from numerous sources to ship one of the best reply.
Take Dow Jones’ Factiva, which now operates an AI-licensed content material market for enterprises. Over the past two years, it has been quietly increasing its licensing agreements for generative AI utilization, with roughly 5,000 publishers from its 30,000-strong writer community now opted in for this.
That was a prolonged endeavor, however one that’s setting the unit up as an efficient negotiator of AI licensing rights, to unlock royalties for small and mid-size publishers on a utilization foundation, in accordance with Factiva’s common supervisor Emma O’Brian, who was employed within the function in August.
In fact, Dow Jones can also be house to The Wall Avenue Journal (each underneath mum or dad Information Corp), a truth O’Brian confused is crucial in figuring out negotiate one of the best phrases on behalf of different publishers. “We might by no means signal a [AI licensing] deal we wouldn’t ourselves signal,” she stated.
That partly means upfront conversations with enterprise purchasers round how precisely they plan to make use of the content material as soon as it’s been RAG’ed, to make sure there is no such thing as a future threat of information leakage, and the way many individuals will likely be utilizing it.
For instance, a shopper would wish to specify in the event that they plan to make use of the content material for his or her inner workforce’s wants, or in the event that they wish to use it to generate a analysis report — a separate use case. Then the Factiva staff will work with the tech departments or information ingest departments of its company purchasers as soon as publishers are onboarded to see the way it works in observe and test that there aren’t any dangers of information leaks. The staff will then test again periodically to make sure they’re upholding their aspect of the contract, in accordance with O’Brian.
Factiva has a utilization API to make sure that any utilization is attributable and that it might probably guarantee its publishers are paid precisely (on a per utilization foundation per RAG question), she added.
Certainly one of O’Brian’s priorities is to make sure specialist, area of interest B2B media firms are additionally getting royalties, not simply the bigger ones, and that Factiva is offering the form of specialist content material wanted for particular industries — relatively than aiming for extra common use.
“I don’t wish to name it a moat, however when you consider rubbish in rubbish out [noisy scraped data] to broadly scraping fashions versus the gold ore that we’ve got in trusted, dependable content material to permit companies to RAG in opposition to, that’s an enormous alternative,” stated O’Brian. “And that’s one thing that may’t be scraped, it must be ring fenced.”
Eye to the longer term: Factiva’s function as negotiator with client LLMs
With hundreds of publishers underneath license and direct relationships throughout its community, Factiva is uniquely positioned to behave as a collective licensor — aggregating rights at scale to barter B2C offers with main LLM platforms. That has the potential to show scattered archives into royalty streams for publishers who’d by no means land these talks solo. That’s an avenue they intend to actively discover, confused O’Brian.
To be clear, it’s in kicking-the-tires mode at this time — the true motion at this time remains to be squarely on B2B: prying open usage-based royalties for publishers through enterprise purchasers’ personal LLMs. But it surely’s not blind to the potential negotiating energy it may yield if it bought the go-ahead from its community to barter with the consumer-facing LLMs.
“The existential query is: ‘how are our publishers and the way are we managing the relationships with the LLMs straight?,’” stated O’Brian. “That’s a client query, and that’s one thing that we hope to have an even bigger hand in down the highway. We wish to be on the desk and have conversations with writer companions on how we may help with that.”
The rise of enterprise LLMs is opening the door to new AI licensing alternatives for extra common curiosity client titles additionally like The Economist and The Monetary Instances. In the meantime, different collective licensing and rights our bodies are additionally shifting in the identical course — carving out a promising new stream of income for the lengthy tail of publishers.
Opens up AI royalties for long-tail publishers
Enterprise urge for food for vetted, high quality content material to feed personal LLMs, combined with the rise in small language fashions, are offering a beacon of hope for the lengthy tail of publishers, usually minimize out of any bigger LLM licensing negotiations, confused Tom West, CEO of Publishers’ Licensing Providers (PLS), a not-for-profit collective administration group representing U.Ok. publishers in collective licensing.
He stated that there’s an emergence of small LLMs specializing in particular industries and matters, for instance, horticulture, and due to this fact are on the hunt for high quality specialist content material for area of interest publishers.
And since enterprises worth provenance, compliance and audit trails, vetted journalism and evergreen explainers turn into premium inputs, with clear attribution and tighter management than client AI permits.
As extra firms use AI to do analysis, analyze information, summarize articles, and deal with admin work, an enormous hurdle is ensuring the fabric they feed these instruments is correctly licensed and used inside its rights. But solely 62 p.c of staff understand their firm’s copyright coverage additionally covers utilizing third-party content material with AI, per the Copyright Clearance Heart’s (CCC) 2025 Content material Utilization Tendencies research.
Smaller publishers can faucet this demand by way of aggregators or vertical partnerships, whereas bigger teams can strike direct, data-rich offers. It’s not flashy, but it surely’s sturdy — and it turns editorial again catalogs into recurring, enterprise income.
West stated that there is no such thing as a technique to calculate presently how a lot significant income publishers could make this manner, given how early it’s and the way it hasn’t but accomplished a licensing template for all content material. However presently, it pays out over £50 million ($66.5 million) a yr to publishers from collective licensing and half of that comes from company licensing. “I’m actually excited by the potential for us to curate content material units for these smaller organizations, bundle that collectively after which license it to start-ups [creating SLMs]. That’s a totally untapped market,” he stated.
PLS, which is owned by U.Ok. commerce our bodies together with the Skilled Publishers Affiliation, works alongside the Copyright Licensing Company, handles copying licenses and returns the cash to publishers.
Within the U.S., the CCC has an analogous roadmap. It has tailored its copyright licensing agreements to incorporate AI utilization, which incorporates company enterprises’ personal LLMs, in accordance with Lauren Tullock, vp and managing director on the CCC. The licenses are opt-in for rights holders and based mostly on utilization information collected by way of surveys. CCC has hundreds of publishers and a whole lot extra for AI rights. The royalties publishers obtain will likely be decided based mostly on utilization, which the CCC calculates by accumulating utilization samples and pays out each six months.
“It’s a really optimistic factor for the small to mid-sized publishers, the place their content material is extraordinarily worthwhile to sure sectors and it’s very tough for purchasers to handle doing direct preparations with each single a kind of,” she stated.
