Introduction

Librarians who work at libraries open to the public (especially those who work at public libraries, public law libraries, and academic libraries that are open to the public) are likely to encounter a patron who is seeking legal advice. I have written about resources that librarians can provide in these situations to provide legal information and not legal advice in Accessing Justice: Supporting patrons with legal information needs.¹ Indeed, research guides are a resource that can help those seeking legal information and Judge Micah W. J. Smith of the United States District Court for the District of Hawai’i has noted that a guide such as the Guide to Virginia Legal Information Resources² is helpful for those seeking legal information.³ However, what guidance can librarians give legal information seekers or self-represented litigants regarding the use of generative AI tools in their workflow? Read on to find out.

The Three Questions Librarians Can Encourage Self-Represented Litigants to Ask

Self-represented litigants may view a generative AI tool as ideal for preparing the first draft of legal pleadings, for example the first draft of a complaint. Librarians can encourage self-represented litigants who use a generative AI tool for drafting pleadings to ask:

1. Is the tool designed for drafting legal pleadings?

Resources for answering this question include but are not limited to the terms of service at the official website of the company that manages the AI tool and Terms of Service Didn’t Read, a project that has its origin in the Chaos Communication Camp held in Germany in 2011 and evaluates how web platforms treat users’ information.⁴ Users should check the terms of service and see what the tool does with input (whether the input is uploaded files, typed content, etc.) Does the tool train on this input? Will buying a subscription to a higher level of service from the tool protect the users’ data from being used as training material? It should be common knowledge that people should not seek legal advice from generative AI tools, however, not everyone understands the purpose of generative AI tools. Terms of service and terms of use, if read by users, can help users of these tools understand what the tools are designed for. As a matter of fact, OpenAI does not condone the use of its tools for making legal decisions that could affect people’s lives.⁵

2. Is the output accurate?

Answer this question by reading the output and checking its citations. Resources for checking citations are the official websites of legislatures, courts, and agencies that contain the full text of statutes, judicial opinions, and regulations.⁶ Self-represented litigants may think of the generative AI tool as something that can check their work, proofread their pleadings and contracts, etc. This may be. Librarians can remind library patrons who are seeking legal information that they should look into what the tool was designed to do. Librarians can recommend resources such as Emily Pavuluri’s Five Editing Lenses Checklist⁷ and AI Critique Review Checklist⁸ for evaluating any feedback that an AI tool provides on a memo, brief, or other document.

Also, librarians can encourage self-represented litigants and other patrons conducting legal research to think about how to write effective prompts. A self-represented litigant may view a generative AI tool as a resource for research. Generative AI tools can be used to create search strings and gather information from other sources. Often, carefully crafted prompts will yield more useful output. Librarians can encourage patrons to ask:

3. How can I craft an effective prompt?

Librarians can recommend resources that help patrons answer this question, such as Korin Munsterman’s CORRECT framework,⁹ Emily Pavuluri’s Responsible AI Use Planning Guide,¹⁰ and the American Association of Law Libraries’ Best Practices for Using AI in Legal Research.¹¹ Directing library patrons to specific tools on prompting can give both clarification and insight to patrons who need assistance in developing their own prompts.

Last but not least, librarians should inform researchers that traditional research tools such as legal treatises, legal encyclopedias, and research guides have not gone out of style. Researchers do not have to do their first search with the assistance of a generative AI tool and some people who work in the field of access to justice recommend that researchers start by searching for sources within a database.¹² The so-called fear of missing out should not guide research decisions; goals should guide research decisions.

References

1. Latia Ward, Accessing Justice: Supporting patrons with legal information needs, Am. Libr. Mag. (June 1, 2022), https://americanlibrariesmagazine.org/2022/06/01/accessing-justice/ [https://archive.is/NmKRn]. In a book chapter I co-authored with Rebecca Katz, I speak in more detail about the challenges and opportunities of generative AI use. See Latia Ward & Rebecca Katz, Access to Justice, in Law Librarianship Practice 266-278 (Ellyssa Kroski ed., 2025). ↩︎
2. Virginia Association of Law Libraries Access to Justice Committee, Guide to Virginia Legal Information Resources: Sources of Law, Research Guides, and Self-Help Books for Self-Represented Litigants (2025), virginia_legal_information_resources.pdf [https://web.archive.org/web/20250423135001/https://www.vacourts.gov/static/courtadmin/library/virginia_legal_information_resources.pdf]. ↩︎
3. Judge Micah W. J. Smith, United States District Court for the District of Hawai’i, How Connection Creates Leadership – Power, People, and Purpose, Keynote Address at the American Association of Law Libraries Annual Meeting and Conference (July 18, 2026) (outline available in the American Association of Law Libraries eLearning platform). ↩︎
4. ToS;DR, About ToS;DR (no date), https://tosdr.org/en/about [https://archive.is/wfeWi]; Chaos communication Camp, ToS;DR, (2011), https://events.ccc.de/camp/2011/wiki/ToS;DR [https://archive.ph/fhEt6]. ↩︎
5. “You must not use any Output relating to a person for any purpose that could have a legal or material impact on that person, such as making credit, educational, employment, housing, insurance, legal, medical, or other important decisions about them.” OpenAI, Terms of Use (Jan. 1, 2026), https://openai.com/policies/row-terms-of-use/ [https://archive.is/YQIQA]. ↩︎
6. For example, govinfo.gov (which contains the United States Code and United States Courts opinions), federalregister.gov (which contains proposed rules, final rules, and more), ecfr.gov (which contains the Code of Federal Regulations), and the Cornell Legal Information Institute which provides a listing by jurisdiction of the websites of branches of government throughout the United Sates. ↩︎
7. Emily Pavuluri, Five Editing Lenses Checklist, (June 1, 2026), https://ailegalresearch.org/downloads/five-editing-lenses-checklist.pdf [https://archive.is/Qof05]. ↩︎
8. Emily Pavuluri, AI Critique Review Checklist, (June 1, 2026), https://ailegalresearch.org/downloads/ai-critique-review-checklist.pdf [https://archive.is/IMTRS]. ↩︎
9. Korin Munsterman, GenAI in Legal Education: A Practical Guide for Professors and Students 106-113 (2026), https://www.cali.org/books/genai-legal-education-practical-guide-professors-and-students. A digital version of the book is available free of charge at the website of The Center for Computer-Assisted Legal Instruction. ↩︎
10. Emily Pavuluri, Responsible AI Use Planning Guide, (June 1, 2026), https://ailegalresearch.org/downloads/responsible-ai-use-planning-guide.pdf [https://archive.is/Q8bXq]. ↩︎
11. Nicole Belbin, Sara V. Pixon, Laura Moorer, Amy Small, & Legal Information Services to the Public (LISP) Special Interest Section, American Association of Law Libraries, Best Practices for Using AI in Legal Research: A Guide for Non-Lawyers, (2026), https://www.aallnet.org/lispsis/wp-content/uploads/sites/11/2026/09/Best-Practices-for-Using-AI-in-Legal-Research-A-Guide-for-Non-Lawyers.pdf [https://archive.is/NHtQZ]. ↩︎
12. Demers notes that tools such as ChatGPT are not permitted to scrape CanLII so get the law directly from CanLII and she also notes that CanLII also has an AI tool (at about the 52-minute-mark). Grossman agrees with Demers’ recommendation to begin research with CanLII and not an AI tool such as ChatGPT (at the one-hour-and-18-minute-mark). Even though the webinar was in the context of searching for Canadian Law, the principle of starting with an authoritative database that provide the full text of the law applies across jurisdictions. For example, people searching for statutes on a certain topic could begin their research at uscode.house.gov by browsing the Titles of the United States Code or doing a key word search there. National Self-Represented Litigants Project, Maura R. Grossman (a professor at the University of Waterloo), Annette Demers (a law librarian and Chair of the Canadian Association of Law Libraries AI Working Group), and Lasandra Conliffe (an educational developer at Ontario Tech University), YouTube, AI, Equity, and Access to Justice: Practical Guidance for SRLs and Legal Professionals (Mar. 25, 2026), https://www.youtube.com/watch?v=AdximqF_PXs. ↩︎