(28 May 2026) In a December 2025 statement, the International Committee of the Red Cross expressed concern that AI tools were producing believable archival and legal references when information did not exist to support them, simply because their inherent purpose is to provide answers. LLMs “will invent details that appear plausible but have no basis in the archival record or legal sources.”
The information science community is one of many groups anxious about AI – it has the potential to and has created hallucinated references to research that has not been conducted, and can be subsequently used in the actual process of archiving by making choices about what to discard and what to keep.
We asked Neil Davies, Senior Manager of Content Acquisition at Durham University, what his experience of AI has been at the university library:
“There are worries that the use of AI has already led to a drop in usage of subscribed materials, where AIs just “scrape” the abstracts (or even the full text) of academic articles and put together a summary, meaning students no longer actually click on articles, and a use is thus not registered by COUNTER or other statistical systems. In this case it becomes harder to know what’s actually being used, which is a concern.”
In this context, how has the importance of preserved content changed with the recent acceleration of AI?
Read more here.
