Artificial Intelligence (AI) in Collection Development

AI in Library Collections

With the increase in the use of artificial intelligence to create content, AI-generated and AI-assisted materials are making their way into library collections, both intentionally and unintentionally. Library staff should stay informed about this issue to understand the implications of AI materials in the collection.

Let’s start with some definitions and examples:

AI-Generated Materials: Content created entirely or partially by artificial intelligence using generative AI tools (such as ChatGPT or others). The content has minimal human input, such as a prompt. Examples include:

Text, images, sounds, or videos entirely generated by AI
Materials containing portions of content entirely generated by AI

AI-Assisted Materials: Content created by humans where the creator has used AI tools to edit, refine, or error-check the work. The human-created content is enhanced or improved by AI tools. Examples include:

Text written by a human and edited or refined using an AI tool
An audiobook written by a human and voiced by AI
An image or artwork created by a human and enhanced with AI tools

AI Hallucinations: Incorrect or misleading results generated by AI tools

Why should libraries pay attention?

AI is having a major impact in libraries by creating challenges in selecting quality library materials for the collection. These materials have the potential to disrupt the integrity of the library’s mission, values, and credibility as a trusted resource. AI-generated materials may contain:

Poorly written text or poor-quality images
Incorrect or misleading information as a result of AI hallucinations – this is especially problematic for nonfiction materials
Falsified or undisclosed authorship
Failure of acknowledgment of the use of AI tools – materials may be misrepresented as human created content

How to identify AI generated materials:

Pima County Public Library provides some helpful tips:

Check the cover image: Are words misspelled or does the image look off? Does the cover not match the content of the book? These errors indicate AI.
Search the author: Most contemporary authors have a website or some kind of online presence. Check also for legitimate reviews or mentions in book-group chats.
Language clues: Look for red flags in the writing, such as sentences that are all the same length, or language that is repetitive or seems unnatural. Check for errors in grammar, and you may even see gibberish. A recipe in an AI-generated cookbook may be missing ingredients or proportions may look off.
Read the online reviews: Does anyone else suspect the book is AI Generated? This is a clue. Look for reviews but be aware that reviews can also be generated by AI so check for “verified purchase” signs. Is the reviewer’s ID just a string of letters and/or numbers? Then it may be a bot.
Think critically and seek out reliable sources: if you have any doubts or questions, go for a book that is proven to be legitimate, such as those published by a reliable and established publisher, in a reliable series, or by a well-known author.
Beware of phony companion books: AI can almost instantly spit out so-called companion books, such as summaries or guides that capitalize on a popular human-written work. These can appear within 24 hours of the original work’s release. These books do not provide original analysis, but rather repeat a book’s conclusions, confusing buyers purposefully and violating copyright.

Wikipedia also provides a handy guide to identify signs of AI writing

For text-based materials, try running the text through an AI detection tool, such as Turnitin. Note that these tools may not always detect with 100% accuracy but may provide some additional clues.

AI in Collection Development Policies

Your library’s Collection Development Policy should include language addressing the library’s stance on AI generated materials. Policies can vary, some may completely prohibit AI content, some may offer some middle ground (with a distinction between AI-generated vs. AI-assisted content), or some may allow with designation. However your library chooses to address this issue locally, IFLS recommends aligning your policy with MORE Collection Maintenance Standards.

See IFLS’s Collection Development Policy example, updated in 2026 with a new section on Artificial Intelligence.

Other examples:

What should you do if you find suspected AI materials in the library collection or catalog?

Library staff should report suspected AI-generated materials not clearly described as such in the catalog record via the IFLS help desk. Providing a title and bibliographic record number(s) is helpful.
IFLS staff will research the title in question using recommended identification methods.
IFLS staff may provide a recommendation to the owning library(ies) and/or cataloging staff based on their findings and in alignment with MORE Collection Maintenance Standards.
If IFLS staff find that the title in question likely contains AI-generated content and the owning library chooses to retain the material, IFLS staff will indicate suspected AI-content in a brief catalog record.

NOTE for e-materials – Suspected AI materials in Wisconsin’s Digital Library (OverDrive/Libby) should be reported in a similar manner (by submitting a help desk ticket). IFLS staff will address reports as guided by WPLC policy/procedure.