Texas School Districts are Overwhelmed by New Library Laws: AI is Not the Answer

{Guest Blog Post by Rachael Welsh} In May of this year, several newspapers, including the Chicago Sun-Times and the Philadelphia Inquirer, published a list of books to read over the summer. While this would normally be an eventful article, a list of books for vacations and poolside reading, there was a problem with this list: ten of the fifteen books weren’t real. 

The list had been generated by artificial intelligence, a phrase that’s becoming more and more commonplace as society figures out the role of AI generators and software. In education, we are inundated with claims that AI can make our jobs easier by helping with creating lessons, making assignments, and even grading, while also making our jobs harder as students turn in work that clearly was not written by a person.

As the 2025-2026 school year began, so did the laws and regulations from the 89th Legislature of Texas. Public schools in particular were hit heavily this session with over 250 laws pertaining to public education; school libraries have been in the headlines as school administrators, school boards, and school librarians interpret Senate Bill 13. This bill requires any books not existing in a school library catalog to be on a list for public community comment for 30 days before going to the school board for approval. Districts also have the opportunity (or are mandated, dependent upon their community) to create a School Library Advisory Council, or SLAC, composed of community members appointed by the school board members and school personnel (who are non-voting members). 

It’s a lot. It’s a lot for librarians for sure, but also for the school board members who are now approving lists of new books that can have hundreds of books submitted by librarians eager to get new books for their students. School board members are elected by the community, unpaid, and do not have any particular training in education. 

Unfortunately, that’s where AI seems to be stepping in at many school districts across Texas. School administrators are using AI generation software to create summaries of the books, to analyze titles for “violations” of the bills (including Senate Bill 12 which doesn’t pertain to library materials), and to examine titles that exist in the library catalogs already. 

Going over book lists is a daunting task, especially when you have multiple schools combining lists and submitting them to a school board with a 30 day turn around. School boards and school administrators aren’t trained in collection development – they don’t know where to find book reviews, what teachers are doing lessons that require text resources, or what topics and authors are popular. Using an AI model can be very tempting, as it also allows for administrators and board members to be able to point to an “objective” source; however, turning to AI to analyze lists is not the answer! As the chair of the TxASL Legislative and Advocacy Committee, we have heard stories from librarians across the state as to issues that are coming up when administrators run their book lists through an AI mechanism for analysis. One district came up with lists of “Potential SB13 Violations” (though SB13 doesn’t have any content guidelines). The books on the potential violation lists included a biography of Wilma Mankiller, the new “I Survived” book, and a new middle grade mystery called “Death by Whoopie Cushion”. Leander ISD was recently in the news due to school administrators running their curriculum books for units of study through an AI in response to SB12, and the books that were targeted there included such classics as “Les Miserables”, “To Kill a Mockingbird”, and “The House on Mango Street”. Students are losing access to books that are windows, mirrors, and doors and are full of vital information and experiences for their life in a multicultural society. 

Using AI to analyze these book lists and library collections will continue to result in these sorts of issues. AI isn’t able to look into thematic elements of books, relying solely on headlines or data that has been previously input into an AI system. AI can hallucinate book titles or information about books, or become biased about titles or subjects based on previously added data – yes, AI can become biased or prejudiced. The AI system can also incorrectly flag subjects as inappropriate or explicit; a book that is about a character’s “sexual awakening” as phrased in a review could be about a character’s first kiss. 

Hopefully districts that are finding their way through the murky waters of these bills will learn from these districts and will consult their librarians and curriculum directors in the future. We are the ones that know the books, the collections, the curriculum, and our students’ and teachers’ needs. To paraphrase Ken in “Barbie”, “our job is books” – and we’re very good at it! 

About the Author

Rachael Welsh has been a school librarian in Texas since 2014. She is the current chair of the Texas Association of School Librarians’ Legislative & Advocacy Committee and is the chair-elect for the Texas Library Association’s District 8. Follow Rachael on Instagram at @rachaelreads