They don’t go missing or get torn and tattered, but e-books are posing concerns for US libraries as publishers insist on restrictive and costly digital licensing contracts, librarians say.“We have to pay for every single checkout, have major limitations on how many copies we can have... and a lot of other arbitrary issues,” said Alison Macrina, a librarian and director of the Library Freedom Project, an advocacy group.Digital collections — including e-books, audiobooks, music and more — have become increasingly central to libraries’ work, particularly since the Covid-19 pandemic when they allowed lending to continue during lockdowns.Patrons checked out a record 662mn e-books and other digital products from libraries last year, 19% more than the previous year, according to OverDrive, a major platform.Over the past decade-and-a-half, the handful of companies that control most US e-book production and distribution have started to lease these works to libraries — rather than selling copies outright.Dubbed “the Netflix model” by some librarians, licensing is not only more expensive, but some worry it allows companies to track reading habits, remove books or censor content.“Major publishers offer no option for the vast majority of e-books to be owned at all by a consumer, whether an individual or a library. You buy a license to view the file,” said Lia Holland from Fight for the Future, a digital rights non-profit.The clash of interests between publishers and libraries has resulted in a series of legal battle in recent years.Publishing companies worry that constraints on e-book licensing could hurt the sector’s economics, while libraries argue the higher fees and other restrictions undermine their mission to make books easily available and encourage reading.“It’s an illustration of the vehemence of this push toward profit maximisation at the cost of an educated populace,” said Holland, campaigns and communications director at Fight for the Future, which has been meeting federal lawmakers on the issue.A number of states have considered laws to oblige publishers to make e-books available to libraries on “reasonable terms”. But publishers and authors have warned the proposals would lower the value of literary works, and a federal judge in 2022 ruled one such state law in Maryland was unconstitutional.Two copyright lawsuits now threaten further restrictions to how libraries can make digital works available.In 2020, four major publishers sued the Internet Archive, a non-profit library with some 44mn print materials and also the world’s largest archive of the Internet.The publishers seek to limit what is known as controlled digital lending — the library’s ability to purchase a book, scan it and then lend the digital copy.Music publishers also brought a second lawsuit over some of the group’s audio recordings.“It’s about ownership — library ownership versus licensing — and the tension that exists between those two ways of managing materials,” said Chris Freeland, director of library services at the Internet Archive.Freeland said the issue was crucial for reader access as well as preservation: “We can’t preserve what we don’t own.” Terrence Hart, the general counsel for an industry trade organisation, the Association of American Publishers, said last year the “Internet Archive’s industrial scale format-shifting activities constitute copyright infringement.” “There is simply no legal support for the notion that Internet Archive or a library may convert millions of e-books from print books for public distribution without the consent of, or compensation to, the authors and publishers,” he said.A judge sided with the publishers last year, but the Internet Archive appealed and the case is ongoing.Longstanding fights over content ownership have expanded to control of distribution channels, said Dave Hansen, executive director of Authors Alliance, which represents authors and submitted a brief in the Internet Archive lawsuit.He said there were now four major e-book publishers in the US, each with their own rules.“These private contracts, private terms and private pieces of technology have supplanted the more generally applicable rules that we have under copyright,” he said.Hansen pointed to a 2022 incident when publisher John Wiley and Sons suddenly removed 1,380 titles from a collection of academic e-books that many libraries use.The experience “demonstrated the power that publishers had to unilaterally dictate what kind of content users could get access to,” he said.Wiley later reversed the decision. It said in a statement it was committed to providing students with affordable e-books and expanding the range of titles available. New technologies are also being used by school boards in their efforts to comply with recently passed state laws banning material that lawmakers have ruled to be offensive.School book bans have increased substantially in recent years and become more comprehensive, according to PEN America, which tracked 5,894 efforts in 41 states from 2021 to 2023.In Iowa, Mason City Community Schools used artificial intelligence (AI) to analyse book content to ensure compliance with a state law passed last year requiring the removal of works depicting sexual acts.“With thousands of books to manage across nine building-level libraries, AI was a tool to efficiently narrow down the list of potential non-compliant books,” Mason City schools superintendent Pat Hamilton said in an e-mail.In December, a federal judge blocked implementation of the state law, pending a legal challenge.Yet the new use of this technology echoes past lessons around AI and social media moderation, said Emile Ayoub, counsel in the liberty and national security programme with the Brennan Center for Justice, a think-tank.“Again and again we’ve seen the limitations of these tools — they’re unreliable, unable to understand content and nuance, they’re biased and can disproportionately impact minority communities,” he said.A tool such as Chat GPT — the “generative” AI program released a year ago — offers a veneer of objectivity, even while producing inconsistent results, he said.“Broad and vague book bans like in Iowa are a basic threat to free speech,” Ayoub said. “And when you use generative AI tools to comply with those bans, it only increases that risk.” Researchers with the Harvard Library Innovation Lab last year tested several “large language models” (LLMs) that power tools such as Chat GPT — asking the models to provide justifications to ban particular books.They found that safeguards against harmful requests were “unpredictable”, with models often going back and forth but complying 75% of the time.“The key point to remember here is that variability is not a bug, but a feature of LLMs,” the lab’s Matteo Cargnelutti and Kristi Mukk said in e-mail.“For now you’ll learn a lot more by talking to a real librarian,” added the lab’s director, Jack Cushman. — Reuters
