Publisher efforts to push libraries to accept AI restrictions appear to be increasingly widespread across the U.S. and Canada. SPARC’s recent webcast on the issue highlighted both the increasing prevalence of these terms and libraries’ ability to push back on them successfully.
Harvard’s Kyle K. Courtney addressed revised copyright statements and the continuing viability of fair use justifications for text and data mining. Key points from Kyle’s presentation are also covered in his recent briefing for SPARC: All TDM & AI Rights Reserved? Fair Use & Evolving Publisher Copyright Statements.
Teri Oaks Gallaway, executive director of SCELC, explained the newly released guiding principles for AI clauses produced by the International Coalition of Library Consortia (ICOLC).
Katie Zimmerman, director of copyright strategy at MIT, provided guidance based on MIT’s experience in weakening AI restrictions, and Anne-Marie Deitering, Oregon State University’s dean of libraries, shared their success pushing back on these restrictions in a recent round of negotiations.
Key Takeaways from the Webcast
- Attempts by publishers to insert AI restrictions into licensing contracts appear to be widespread across the U.S. and Canada and across publishers.
- There is a strong fair use case in the U.S. for academic uses of AI on licensed content, and libraries should try to avoid restrictions (specifically related to AI and more generally) that would prevent users from exercising fair use.
- In addition to planned negotiations, some publishers are seeking to insert new terms regarding AI restrictions mid-contract. Libraries have successfully declined these mid-contract amendments.
- Language restricting the use of AI may be buried in the middle of other terms in vendor agreements. Libraries may want to search contracts for words related to these restrictions to ensure they aren’t agreed to unknowingly. Terms to search for might include artificial intelligence, machine learning, algorithmic development, and deep learning.
- When libraries raise concerns about AI restrictions, some publishers have pushed back by saying that other customers aren’t objecting. Regardless of the veracity of this claim at one point in time, it is clear that libraries are deeply concerned about these terms and an increasing number are or are planning to push back on them strenuously.
- Panelists suggest addressing the company’s primary concerns (unauthorized access, using content to train products that could compete with the vendor) to find common ground while emphasizing the need to preserve libraries’ rights and interests in providing access for academic use of AI and TDM tools and avoiding liability for restrictions that are unenforceable. Existing contract terms will likely already address vendors’ primary areas of concern, and institutions can seek carve outs for existing permitted uses (text mining, uses incidental to permitted uses, de minimis uses).
- Libraries can point to ICOLC’s Statement on AI in Licensing and the UC Office of Scholarly Communications’ recent analysis, “Fair use rights to conduct text and data mining and use artificial intelligence tools are essential for UC research and teaching,” as validators for the importance of maintaining rights necessary for academic uses of AI.
- Resources are available for helping libraries develop their own language for removing or weakening AI restrictions.
- The UC Office of Scholarly Communications has adaptable licensing language that addresses publisher concerns while retaining users’ ability to use AI tools.
- ICOLC is supporting member consortia in negotiating these terms. Consortia your institution is a member of may be able to provide additional support.
- Resources like the SPARC Contracts Library and ESAC Registry will likely have a growing number of examples of negotiated language (such as Oregon State’s most recent Elsevier contract).
- As more institutions push back on these terms, the negotiating position is likely to improve for everyone.
The recording of the webcast is limited to SPARC members only. SPARC members will receive an email with instructions for accessing the recording and, if needed, may request access to the recording by emailing [email protected].
The webcast clearly demonstrated the growing momentum among libraries to counter publisher AI restrictions. SPARC is committed to helping members navigate this rapidly evolving area of concern.
For institutions whose negotiations are not covered by NDAs or confidentiality agreements, we would ask that you share your experience with SPARC to enable us to track libraries’ evolving experiences in this area. You can share your experience through the form here or by emailing [email protected].