The UC Berkeley School of Law (Berkeley Law) AI Policy is a formal, institution-wide policy effective Summer 2026. As written, it defaults to prohibition of AI assistance across nearly all course work rather than to disclosure-and-consent. It is among the first widely-circulated U.S. top-tier law-school AI policy texts to adopt a prohibition default; most U.S. law-school AI policies through 2025 instead defaulted to case-by-case instructor discretion with optional disclosure, the model that has predominated since 2023.
Default rule
The policy prohibits the use of AI for aid in conceptualizing, outlining, drafting, revising, translating, or editing any work submitted for credit. It prohibits AI for "any use for any purpose in any exam situation." Students may not upload course materials — assignments, readings, slides, class recordings, or other class content — into generative AI systems.
The policy carries one narrow exception: AI can be used for research on papers only for the limited purpose of identifying sources, such as cases, statutes, or secondary sources. Students remain responsible for the accuracy of their research and all other aspects of their submitted work.
On enforcement, the policy states that "Citations to sources that do not exist will raise a presumption of prohibited AI use." This converts the hallucinated-citation pattern (cf. Sycophancy and Hallucination and the AI content saturation / pro-se filing fake-citation surge) into evidentiary grounds for an honor-code presumption. The fake-citation pattern in legal filings was documented in The Washington Post in May 2026. Most law-school AI policies do not explicitly enumerate translation as prohibited; Berkeley's explicit ban on translation, and its course-material upload prohibition, are stricter than is typical, with the upload prohibition addressing training-data and copyright concerns about student-uploaded course materials entering generative systems.
Activities enumerated as violations
The policy lists specific examples of prohibited use:
- Asking an AI tool to brainstorm a paper topic or thesis (prohibited conceptualizing).
- Asking an AI tool to propose an organizational structure for a paper (prohibited outlining).
- Asking an AI tool to compose a paragraph summarizing a legal rule (prohibited drafting).
- Asking an AI tool to identify repetitive passages that should be cut (prohibited revising).
- Asking an AI tool to polish a paper by correcting grammatical mistakes (prohibited editing).
- Asking AI to generate an exam outline whose elements are then used on the exam (prohibited exam use).
- Asking AI to translate a paper originally written in another language into English (prohibited translating).
Stated rationale
Berkeley Law frames the policy as protecting the cognitive-skills core of legal education against premature AI substitution. The policy states that "Thinking remains the sine qua non of good lawyering (and of a quality legal education)," and is structured to ensure courses focus on requisite cognitive skills by default, providing students "the opportunity to develop the skills they need to conceptualize, outline, draft, revise, and edit their work." The translation prohibition is framed as a means to develop legal-English fluency for ESL students.
The policy states two purposes: (1) "the best legal education possible… by equipping [students] to perform activities constitutive of excellent lawyering, such as mastering primary texts, using legal reasoning to apply legal authorities to novel legal questions, and independently developing creative solutions"; and (2) fairness and administrability, the latter reflecting that the disclosure-and-consent model has proven hard to administer in mass classes.
Discretionary deviations
Instructors may deviate from the default rule, but only "in writing and with appropriate notice," and must "require students to disclose any authorized AI use." AI-fluency courses are explicitly contemplated as a category where deviation is appropriate; the policy thus treats AI fluency as a separate course track rather than a competency spread across the curriculum. On the student side, the policy provides that "If a student has a question about whether a particular use of AI violates this default rule or an instructor's alternative rule, they must ask their instructor and receive clarification in writing before engaging in the use."
Reception and comparison
Berkeley Law's institutional default of prohibition is distinct from the default-permitted-with-disclosure model that has predominated since 2023, positioning the text as a possible template for other law schools addressing the same questions. The hallucinated-citation presumption operates as the policy's main enforcement lever, converting a documented LLM failure mode (fake legal citations) into a per-se evidentiary presumption of prohibited AI use (cf. Sycophancy and Hallucination, AI Content Saturation (\"AI Slop\")). The explicit translation ban and its legal-English-fluency rationale are positions that ESL-heavy programs will have to engage.
The Vatican's April 2026 framework bans AI use in homily writing, an analogous role-integrity rationale preserving pastoral and professional cognitive labor in a different institution. Higher-education AI policies more broadly remain in flux; cf. AI in Education.
Relationships
- supports: AI in Education, Legal Services — AI Deployment (legal-profession-side parallel to law-school-side concerns), Sycophancy and Hallucination.
- related: Legal Services — AI Deployment, Education — AI Deployment, AI Content Saturation (\"AI Slop\"), AI Fluency Divide.
- contrast: Pope Leo XIV Vatican AI framework (April 24, 2026) — analogous role-integrity rationale in a different institution.
Citation form
Cite as [[sources/uc-berkeley-law-ai-policy-summer-2026]].