University of Michigan-Ann Arbor

U-M sample syllabi allow instructors to specify GenAI disclosure, restrictions, or bans by activity, in alignment with college policy; examples are not automatically binding campus-wide rules.

U-M ITS AI data are not used for model training, but service input, prompt history, usage metadata and logs are processed, including sharing prompt data with cloud vendors. The notice does not establish a uniform retention period.

ITS AI service and Google services have a U.S.-only usage condition under U-M terms (May 2025).

U-M Google Gemini and Gemini Notebook (formerly NotebookLM) exclude sponsored affiliates, alumni and retirees; supported sensitive data vary by approved guide.

OVPR internal peer-review guidance updated August 11, 2025 discourages generative AI reviewing and strongly discourages uploading other researchers’ grant proposals to commercial AI tools.

An LSA student found responsible may submit a written appeal to the Academic Judiciary Committee within two weeks from the date of the resolution letter, challenging responsibility, the college sanction or both. Grounds include serious procedural error, significant previously unavailable evidence or excessive college sanctions. Sanctions remain during appeal. Discrimination/prohibited-behavior claims must also go to ECRT and pause appeal. AJC may reverse responsibility or recommend keeping, reducing or increasing sanctions.

For another school’s student in an LSA course, SAA determines the course integrity violation and instructor course sanction authority; the student’s home school imposes college sanctions, including any recommendations it accepts.

In LSA SAA Resolution, the Coordinator of Academic Integrity or designee decides under the preponderance standard. Students receive allegations/evidence and may respond, present material and ask questions. Students should notify SAA when bringing a support person. The usual Honor Council participant does not advocate for or against the student.

In LSA, course-grade penalties follow the instructor’s syllabus/department policies; SAA may assign college sanctions including probation, reflective essay, suspension or expulsion. These are possible outcomes, not an automatic AI penalty.

LSA instructors may use Instructor Resolution for non-egregious incidents meriting no consequence greater than zero on the assignment; remedial work may also apply. SAA Resolution is appropriate for possible course failure or more severe sanctions, previous responsibility findings, or falsified academic documents. While pending, students are presumed not responsible, may attend class, cannot drop/change grading option, and instructors may not impose penalties/remedial work. A pending end-of-term case may receive NR and prevents course credit/graduation.

ITS does not provide a campus-wide teaching/learning plagiarism checking system. Detector output can trigger inquiry but is not definitive proof, while iThenticate is described for researchers’ own work, not student evaluation.

The School of Nursing faculty guide advises that AI should not be the sole basis for grading and restricts uploads of identifiable student work unless the service and use are approved.

Procurement Services should be the sole purchaser for AI services intended for institutional Moderate, High, or Restricted data; access alone does not confer approval for each sensitive data type.