Since lawyers began relying on AI to draft briefs, legal citation has faced new scrutiny. Traditionally, The Bluebook has been the standard for citation discipline. But its 22nd edition introduced Rule 18.3, the first standardized format for citing generative AI, and it has sparked widespread criticism.
What the Rule Says
Rule 18.3 has three parts:
- 18.3(a): Large language models – requires citing the prompt author, model name/version, full prompt text, submission date, and a PDF of the output.
- 18.3(b): Search results – cites the engine, query, number of results, date, and PDF.
- 18.3(c): AI-generated content – covers non-text output, noting the model used and AI authorship.
Key Criticisms
- Conceptual confusion: Overlap between “LLM output” and “AI-generated content.”
- Unclear purpose: The rule explains how to cite AI but not when citation is appropriate. Critics argue AI is a tool, not an authority, like a research assistant or law clerk.
- Internal inconsistencies: Even The Bluebook’s own examples fail to meet its stated requirements.
- Technical burden: Demands screenshots, PDFs, and formatting beyond many lawyers’ skills.
- Mismatch with practice: AI use is iterative and conversational, making full documentation impractical.
- Ethical risks: Requiring citations may expose confidential client information or attorney work product.
Alternative Views
Some scholars, like Susan Tanner, argue AI should almost never be cited directly. Instead, lawyers should cite the sources AI helps uncover. If AI outputs themselves must be referenced, citations should clarify they are records of a conversation, not authorities.
Implications
Law schools, journals, and practitioners now face uncertainty: when, if ever, to cite AI; how to avoid breaching confidentiality; and how to maintain uniformity. Critics describe Rule 18.3 as “deeply flawed,” “bonkers,” and a fundamental misunderstanding of both AI and the purpose of legal citation.
Source: Law Sites
Leave a Reply