Academic and scientific publishers license journal articles, research books and learning content to AI developers, mostly to buyers they do not name, and two listed publishers report what it earns. Wiley states AI licensing revenue of USD 33.1 million in Research and USD 16.0 million in Learning for the fiscal year to April 30, 2026 (10-K), and Informa, the parent of Taylor & Francis, calls its USD 75m+ of data access revenue for 2024 non-recurring (full-year results).
Collective licenses, research exceptions and author consent decide how much a publisher can license, and some publishers also sue AI developers. Licensing is one route within AI training data licensing, and declining to license is a valid choice.
DataPacts Editorial Team checked every figure here against its primary source on November 9, 2026, and the Mercor listing again on November 10, 2026; the page on how we source, date and check every figure sets out the rules.
What Academic and Scientific Publishers License to AI Developers
Academic publishers license the full text of content they hold rights to, from journal archives to textbooks and course material, for training or for access by AI systems. Wiley describes its first project as a "content rights project for training GenAI large language models", and a buyer listing for publishers asks for long-form text, courses or video in the hundreds of titles or hours. The content classes are:
- journal back files and current articles
- research books and monographs
- textbooks and courses
- society journals published under contract
Scale is set by platforms as much as by single houses: Wiley states in its 10-K for the fiscal year to April 30, 2026 that its Atypon platform hosts approximately 2,000 publishers and societies. Academic publishers are one of the types of who sells data to AI companies, and their rights position is the contract with each author and society.
Disclosed Academic Publisher Deals and What Their Filings Show
Only filings by listed publishers put figures on academic AI licensing, and none of them names the buyer. The table lists the filed items by disclosure date, as of 2026-11-09.
| Date | Publisher | Counterparty | What | Figure (stating party) | Training stated | Source |
|---|---|---|---|---|---|---|
| 2024-06-13 | Wiley | not named ("large tech company") | book and learning content rights for training GenAI LLMs, one-time project | USD 23 million (Wiley FY2024 release) | yes | sec.gov |
| 2024-07 | Taylor & Francis (Informa PLC) | not named ("major AI/Large Language Model buyer"; second partnership H1 2024) | academic content and data access | AI-related revenues USD 75m+ in 2024, "non-recurring" (Informa) | yes ("to help train" LLMs, Informa) | investegate.co.uk (RNS) |
| 2024-09-05 | Wiley | not named ("large tech company") | content rights for training GenAI models | USD 21 million project; Q1 FY2025 USD 16 million (release) vs USD 17 million (presentation) (Wiley) | yes | sec.gov |
| 2026-06-01 | Wiley acquires Emerald Publishing (CIG Emerald Holding LLC) | owners of CIG Emerald Holding LLC | acquisition, not a license; "to strengthen its proprietary content advantage" in AI | preliminary consideration USD 462.7 million (Wiley 10-Q) | n/a | sec.gov |
Wiley: Two Content Rights Projects and a Shift From Learning to Research
Wiley disclosed two GenAI content rights projects in 2024, USD 23 million on June 13 and USD 21 million on September 5, each with a buyer it describes only as a large tech company. See Wiley's AI licensing deals for each project. Wiley's own documents of September 5, 2024 disagree on the second project's first-quarter amount: its release says USD 16 million and its presentation says USD 17 million, and the difference is not resolved in either.
The segment split has moved since. In the quarter ended July 31, 2025, Research AI licensing revenue was USD 16 million against USD 1 million a year earlier (release of September 4, 2025). For the full fiscal year to April 30, 2026, Wiley's 10-K reports Research at USD 33.1 million, up from approximately USD 11 million, and Learning at USD 16.0 million, down from approximately USD 29 million. Wiley's separate total of USD 49 million for that year (release, up 23 %) has a different base and is not added to these values. In the six months to October 31, 2025, USD 15.7 million of AI license revenue related to content licensed from other publishers (10-Q). For the quarter to July 31, 2026, the 10-Q gives USD 13.7 million against USD 28.9 million a year earlier.
Taylor & Francis: Revenue Its Parent Calls Non-Recurring
Informa, the parent of Taylor & Francis, reported AI-related revenues of USD 75m+ for 2024 and described this data access revenue as non-recurring. Its half-year report of July 24, 2024 says a "second major partnership" was secured, after a first with a "major AI/Large Language Model buyer" in its May 2024 trading update, and no buyer or term is named. The deals are covered under Taylor & Francis AI licensing deals.
The label has a visible effect: Informa reports underlying growth of 4.0 % in Academic Markets for the first nine months of 2026 excluding non-recurring data contracts (trading update, as of September 30, 2026). With its half-year results of July 30, 2026, Informa states it is "exploring a range of opportunities both with AI companies and other corporates".
Other Academic Publishers: An Acquisition, Author Permission and No Filed Terms
Most academic publishers have disclosed no AI licensing terms, and the public record for them is limited to an acquisition and a statement by an authors' group. Wiley's purchase of Emerald Publishing on June 1, 2026, for preliminary consideration of USD 462.7 million (10-Q), is an acquisition and not a license, and Wiley states it is meant to strengthen its proprietary content advantage in AI. The Authors Guild named Cambridge University Press on November 19, 2024 as seeking author permission for AI licensing, and no deal is stated. No terms or figures for other academic publishers are on public record.
What AI Buyer Programs Ask Academic Publishers For
One published buyer listing for publishers and education companies asks for long-form content the company has the right to license and excludes content that is already open access. Mercor states a range of USD 35,000 to 370,000 to the company that depends on catalog size and rights (checked November 10, 2026), a program claim and not a paid amount. The listing sets three groups of conditions:
- wanted: rights to license for AI training, long-form text, courses or video, hundreds of titles or hours
- not a fit: content distributed under someone else's license and content already open access
- tools: CMS, LMS and XML or EPUB pipelines
Mercor also states a one-day fast-signing bonus, a reason for a publisher to review rights at its own pace. The listing sits within Mercor's buyer program terms and reputation profile, where the program's other terms and its record are set out.
Collective and Intermediary Routes for Journal and Book Content
Publishers that do not sign direct deals can license through collective schemes, which grant AI developers non-exclusive rights to a pooled repertoire. The table compares the routes published, as of 2026-11-09, with the Mercor row re-read on 2026-11-10.
| Route | Operator | Scope as published | Money terms published | Source |
|---|---|---|---|---|
| Buyer program listing | Mercor (buyer) | "Publisher or Education Company": rights to license, long-form text, courses or video, hundreds of titles or hours; not open access | Mercor states USD 35,000 to 370,000 | api.data.mercor.com, checked 2026-11-10 |
| Collective license (US) | Copyright Clearance Center (collective licensing organization) | AI Systems Training License: voluntary, non-exclusive, training AI for uses external to the licensee; internal AI re-use rights in the Annual Copyright License since July 2024 | not published | copyright.com, checked 2026-11-09 |
| Collective license (UK) | Copyright Licensing Agency (collective licensing organization) | Generative AI Solution: opted-in books, journals, magazines; training, fine-tuning, RAG; publishers opt in via PLS, authors via ALCS | not published | cla.co.uk, checked 2026-11-09 |
| Direct deal | Taylor & Francis; Wiley | one buyer per project, buyer unnamed | aggregate figures only | filings |
CLA states its solution is "designed to work in harmony with any direct deals publishers have or want to pursue", so the two routes can coexist. Neither collective organization publishes fees. Terms and scope of such schemes are compared under collective licensing for AI training.
Copyright Exceptions That Limit What a Publisher Can Charge For
Research exceptions let some users mine scholarly content without a license, so a publisher's licensable value lies mainly in commercial training. In the UK, section 29A of the Copyright, Designs and Patents Act 1988, in force since June 1, 2014, permits copies for text and data analysis for non-commercial research only, and section 29A(5) makes contract terms that prevent it unenforceable.
Under Regulation (EU) 2024/1689, Article 53(1)(d), providers of general-purpose AI models publish a training content summary, and these obligations have applied since August 2, 2025. The template discloses licensed datasets only as Yes or No plus modality, never as a full list.
Under Directive (EU) 2019/790, Article 7(1) protects the Article 3 research exception from contract override, while Article 4, the commercial exception, does not have that protection, and an Article 4(3) reservation for online content must be machine-readable. How this works is set out in the EU text and data mining exception explainer.
Authors, Editors and Consent Before an Academic License
An academic publisher can license only the rights its author and society contracts give it, and authors' groups ask to be asked first. The Authors Guild commended Cambridge University Press on November 19, 2024 for seeking author permission before AI licensing. Four checks come before any license:
- what the publishing agreement grants, and whether it covers AI training
- backlist titles signed before AI clauses existed
- society-owned journals published under contract
- open access licenses chosen by authors
On money, the Authors Guild recommends that authors receive 75 to 85 % of AI licensing income (December 12, 2024), an advocacy position for books and not a rate that any academic publisher applies. No academic publisher has published an author payment rate. Book publishers face the same questions, covered by book publishers and authors licensing books.
Litigation Alongside Licensing: Academic Publishers as Plaintiffs
Academic and educational publishers also sue AI developers over training copies, so licensing and litigation run side by side. Elsevier, Cengage, Hachette, Macmillan, McGraw Hill and Scott Turow sued Meta Platforms and Mark Zuckerberg in a proposed class action alleging willful copyright infringement in training Llama.
The case was filed on May 5, 2026 in the Southern District of New York and transferred on September 25, 2026 to the Northern District of California as 3:26-cv-11415-VC. It is pending with no trial date as of November 9, 2026, according to plaintiffs' counsel.
A running list of rulings and filings sits in the case tracker of AI copyright lawsuits over training data, which follows the docket.
Licensing or Not: Benefits, Risks, Downsides and Alternatives for Academic Publishers
Licensing brings revenue the filings show can be large, but the same filings show it can fall sharply from one period to the next. The lists below set out the trade-offs.
Benefits
- revenue that Wiley reports as USD 49 million for the fiscal year to April 30, 2026
- income from relicensing other publishers' content, as in Wiley's six months to October 31, 2025
- a paid use for backlist content under rights the publisher holds
Risks
- sharp swings, as in Wiley's Q1 FY2027 AI revenue against the prior year (10-Q, quarter to July 31, 2026)
- revenue labeled non-recurring, as Informa labels its 2024 data access revenue
- a quick-signing incentive that can rush the rights review
- litigation by peers over the same kind of training copies
Downsides
- research use under the exceptions stays free
- a training summary under the AI Act shows licensed data only as Yes or No plus modality, so the license adds little visibility
- author consent and society contracts can limit what is licensable
Alternatives
- a collective route that pools the repertoire and works alongside direct deals
- a machine-readable reservation of rights
- licensing a subset of titles only
- not licensing at all
Each route turns on the rights held and the buyer's terms, and the general decision is covered under should you sell or license your data.
How Academic Publishers Fit Among Other Content Sellers
Academic publishers share deal structures with news and book publishers but differ in their buyers' interests and in the research exceptions that apply to them. Deals by news publishers and media licensing content to AI companies are covered separately. The contract and research pages cover adjacent questions:
- AI data licensing agreement clauses
- retraining and model retention rights
- universities and research groups licensing data
- scientific data and papers as AI training data
Related Deal, Revenue and Law Pages
The figures on this page come from filings and organizations' own pages, each listed with its source date. This page is information, not legal advice.
- AI data licensing deals tracker
- company data partnership programs
- AI data licensing revenue in company filings
- data from acquired companies
- UK copyright and AI training
- machine-readable TDM reservations
Frequently Asked Questions
Which academic publishers have licensed content to AI companies?
Wiley and Taylor & Francis are the academic publishers with AI licensing figures in their own filings; neither names its buyers. Wiley discloses two projects of USD 23 million and USD 21 million in 2024, and Informa reports USD 75m+ of non-recurring data access revenue for 2024.
Do academic authors get paid when their publisher licenses to AI?
No academic publisher has published an author payment rate for AI licensing; it depends on each publishing agreement. The Authors Guild recommends 75 to 85 % to authors for books, as its own advocacy position (December 12, 2024).
Is open access research licensed to AI companies?
Open access content is generally outside paid deals: one buyer listing names it as not a fit. Open access content can also fall under the research exception of Directive (EU) 2019/790, Article 3, which contract terms cannot override.
How much do AI companies pay for journal content?
No per-article or per-journal price is public; the only figures are publishers' aggregate project values and revenues. They are Wiley's USD 23 million and USD 21 million projects, Informa's USD 75m+ of 2024 revenue and Mercor's stated program range of USD 35,000 to 370,000 (checked November 10, 2026). Valuation methods are explained under what company data is worth to AI companies.