By Gaëlle Béquet, 19 November 2025
The 2025 Charleston Library Conference highlighted several major trends currently transforming the academic, documentary and publishing ecosystems in the United States of America. Universities are experiencing a significant decline in student enrolment, a trend exacerbated by a fall in the number of international students. This contraction directly impacts libraries, which are facing reduced acquisition budgets and consequently cutting back on subscriptions to journals and reference databases. In this context, publishers’ transformative agreements are receiving less support, particularly when they impose embargoes on open access content availability.
In scientific publishing, there are several signs of accelerating change. Wiley, for example, illustrates this paradoxical dynamic: while researchers are under increasing pressure to publish, libraries are buying less. Wiley is also continuing to grow in Asia, as evidenced by its office in Beijing with 75 employees, and it has recently published recommendations on the use of AI for authors. Many researchers now object to their articles or books being used to train AI models, while publishers are using detection tools to automatically identify content generated by AI in submitted manuscripts. Finally, several speakers raised the possibility of small publishing houses and learned societies disappearing, as they are threatened by increasing concentration in the sector.
The conference also addressed the evolving role of librarians. According to Lorcan Dempsey, proficiency in programming languages such as Python will be essential for effective interaction with the artificial intelligence systems that are becoming standard throughout the document chain.
Another highlight of the discussions was issues related to open access. Works available via OAPEN are widely captured by robots and AI systems that can bypass authentication mechanisms by mimicking human user behaviour. This automated retrieval disregards Creative Commons licences and undermines data protection efforts. Site managers can no longer rely on the declared identity of visitors, leading some to consider a radical scenario in which websites would no longer offer direct access to data and would instead delegate intermediation to AI, which would present the information to end users. At the same time, several participants noted a decline in the use of Google for information searches, with users increasingly turning to conversational AI tools instead.
The issue of scientific integrity was also addressed. Cabells now lists almost 19,000 questionable journals in its database. The term ‘grimpact’ has been proposed to describe the negative effects that biased research based on distorted data or unethical practices can cause.
Regarding infrastructure and tools, there were presentations on advances concerning the Collaborative Collection Lifecycle Project, notably thanks to the CYCLOPS prototype developed by Indexdata and based on a Recommended Practice currently open for public comment by NISO.
Finally, the conference provided an opportunity to discover new tools for libraries. These include digitisation and optical character recognition solutions adapted to non-Latin languages, such as the Spacy.io tool. There was also mention of an inspiring collaboration between Lehigh University and Elsevier, which aims to create a chemistry literature analysis tool to save teachers valuable time when exploring existing literature prior to conducting experiments.