Authors have until the end of August to opt out.
SvetaZi/iStock/Getty Images
Johns Hopkins University Press (JHUP) is the latest academic publisher to announce plans to license its books to train proprietary large language models. According to an email JHUP sent to authors Tuesday, those who want to opt out of the licensing agreement have until Aug. 31 to sign an addendum to their contracts; otherwise their work is fair game.
The move comes as Johns Hopkins University—the nation’s largest spender on university-based research and development—is facing big budget holes created by the Trump administration’s sweeping cuts to federal grants.
“While we do not anticipate huge financial gain for individual books, the cumulative revenue [from LLM licensing deals] would be meaningful for Johns Hopkins University Press and our mission,” read the email sent to authors. “As we anticipate contraction in the higher-education market, these funds can help to sustain our important work as a non-profit publisher.”
While JHUP is not currently operating at a deficit, its executive director, Barbara Kline Pope, said in an email to Inside Higher Ed that the publisher is “exploring how our financial model may need to evolve over the coming years.” Pope did not answer Inside Higher Ed’s specific questions about which company or companies it plans to license book content to, but said that it’s “currently exploring partnerships with both general AI companies and those focused on specialized content and inference models like Retrieval-Augmented Generation,” which can incorporate external information sources to enhance the authority of an LLM’s response.
The press maintains a backlist of about 3,000 titles and publishes roughly 150 new books a year by faculty and other experts in fields such as public health, science, higher education and the humanities. It told authors that they can expect to receive “modest” returns of less than $100 per title per license.
While JHUP did not provide a specific dollar figure for how much revenue it expects to generate from the licensing agreement, some of the biggest scholarly publishers have already proven that there’s money to be made in licensing content to AI companies.
In the two-plus years since generative artificial intelligence tools have gone mainstream, major for-profit academic publishers, including Wiley and Informa (Taylor & Francis’s parent company), have signed agreements with AI companies. While some optimistic authors and observers have said such deals mean well-researched, accurate data will be used to train AI models, others have pushed back. Last summer, authors were outraged after Taylor & Francis failed to notify them before selling their work to Microsoft for $10 million. By the end of 2024, Taylor & Francis reported a $75 million profit as a result of the sale, which boosted its underlying revenue growth from 3 percent to 15 percent in one year, according to Bloomberg.
In addition to JHUP, other nonprofit publishers are jumping on the AI bandwagon—or at least thinking about it. Last year, Oxford University Press confirmed it was working with AI companies to develop LLMs, while the university itself launched a five-year partnership with OpenAI this past spring. Cambridge University Press is still in the process of weighing AI licensing agreements, though it’s also given authors the opportunity to opt out of any future AI-related aggregation efforts. Massachusetts Institute of Technology Press said in November that multiple AI companies have approached about a licensing agreement; it responded by asking authors for their input and has not publicly announced a deal.
In its notice to authors this week, JHUP said it spent the last year weighing the possibility of licensing its works to train LLMs. In addition to potential financial gain, the press explained that it is deciding to move forward now because an LLM licensing agreement would make authors’ work more discoverable by their intended readers, create some guardrails around content use amid increasing concerns that major LLM companies are already scraping pirated versions of JHUP’s book content, and make a stronger legal case that such companies should be required to pay for access to the publisher’s content.
Sharon Ann Murphy, a history and classics professor at Providence College in Rhode Island who signed two contracts with JHUP long before the rise of LLMs, said she was not surprised—but nonetheless upset—by the notice from JHUP, which includes language from the opt-out addendum. It requires authors who don’t want to license their work to acknowledge that in addition to not receiving any AI-related royalties, “the sales and reach of the Work may suffer as a result of or in relation to the fact that Hopkins Press will not exercise AI Rights with respect to the Work.”
Murphy said she interpreted JHUP’s opt-out clause to mean that authors “are agreeing that they’re going to lose revenue because of this and Hopkins has no responsibility to protect us.”
Murphy is also skeptical of JHUP’s claims in its email to authors that if LLMs adopt technologies that credit the sources of AI-generated response, it will give readers the ability “to identify and click through to the original source” and is “the best way to continue to engage with readers and disseminate (authors’) work widely.”
“They’re saying that somehow this will promote our work, but that’s a specious argument. That’s not how AI models work,” Murphy said. “Academic presses are operating on shoestring budgets, but this seems really short-sighted. Academic presses are in the business of creating real knowledge, but AI is in the business of hallucinating and making stuff up.”
Annette Windhorn, a spokesperson for the Association of University Presses, wrote in an email to Inside Higher Ed that she’s not sure just how many academic presses have agreed to license their content to AI companies.
“An internal query to member presses more than a year ago did reveal that a number of presses had been approached by a variety of companies, but almost none were at that time actually considering an agreement and many presses were deferring initial decision points to university counsel,” she wrote. “Our members are following developments closely, but moving with caution in areas that may impact their authors’, their institutions’, or their own rights and responsibilities.”
