CBA Record November-December 2025

neighborhoods don’t get the funding they need and the disinvest ment grows worse. While I recommend this book with its many thought-provok ing stories, I wish there had been a heavier editing hand. It seems interviews were transcribed verbatim, and some folks are more eloquent than others. Also, while I realize this is not a scholarly study, the method for choosing subjects was skewed. The authors posted a request on Instagram asking for “Don’t Go” stories. Over 70 people responded, which they winnowed to 25 to interview. Thus, the subjects were self-selected to discuss this subject.

Still, this book is a valuable contribution to the problem of segregation and offers simple solutions, if only we try them. As Johnson, says, “Laws and policies are important, but if we really want to disrupt segregation, we have to get to know each other. I wanted all of us to realize what we have in common and become curious about our differences. I want us to care about each other.” As the authors note, segregation doesn’t harm just the people who live in disinvested neighborhoods. We all lose out because we miss the chance to meet each other.

THE BLUEBOOK: A Uniform System of Citation

specifically, with detail that can assist the AI model in generating a response. To do this in practice, case-specific information needs to be used for context. Following this rule as it’s written could result in priv ileged information being disclosed in a citation because it was part of the prompt used to guide an attorney’s work. At the very least, the time spent crafting effective prompts is attorney work product that must now be disclosed as part of the cita tion, according to this rule. Additionally, Rule 18.3(b) requires the same disclosure for searches in either web browsers or research services such as Westlaw, Lexis, and others. This rule requires citation of the name of the search engine and “the exact text of the query (including any Boolean operators employed) in quotation marks” in addi tion to the number of results, the date of the search, and a parenthetical disclos ing where the PDF is stored. One of the examples given is “Westlaw, +“Disclosure Controls” +“SEC”, 1 result, (Mar. 30, 2024) (on file with Yale Law Journal) (fil tered by “Cases”, “2d Cir.”).” Based on the way the rule and exam ples are written, searches made as part of an attorney’s research must now be cited along with detailed information that led to the results. As with the prior section, this raises concerns about the disclosure of privileged information and attorney work product for purposes of citation. These aspects of Rule 18.3 raise con cerns about privilege, work product, and privacy as they intersect with genAI and citation.

By Harvard Law Review Reviewed by Meredith Geller, Northwestern Pritzker School of Law and CBA Record Editorial Board Member

T he Bluebook’s 22nd edition was just published this year. As one might imagine, it takes on the hot-button issue of citing to AI-generated content in Rule 18.3. But several aspects of the new rule raise concerns related to the intersection of genAI and citation.

Although 18.3(c) specifies how to cite AI-generated content, other parts of the rule raise concern about its impact on attorney research by requiring disclosure of the research process, as opposed to merely the sources located. Rule 18.3(a) provides that if an attor ney uses a generative AI service such as ChatGPT or even Google search, a screenshot of the search output must be maintained in a PDF: A URL is no longer sufficient, given how data is generated and the number of available web sources. This PDF must remain on file for an undisclosed period. Further, 18.3(a) requires that, “The prompted responses of large language models, such as ChatGPT and CoPilot, should be cited by the author of the prompt, the name of the model used (including version number if the model is identified according to ver sion), the exact text of the prompt submission in quotation marks, the date when the prompt was submitted, and a parenthetical indicating where the PDF is stored…” The example given of how to cite in this way is, “Luke Cronin, Google Gemini Advanced, “Who would make a better Supreme Court Justice: Beyoncé or Taylor Swift?” (Mar. 29, 2024) (on file with the Columbia Law Review). Although the example is amusing, the real-world implications are not. Westlaw and Lexis both have genAI tools as part of their services, which would mean this rule applies to research done using these tools. Effective prompts to get results from genAI must be written

CBA RECORD 39