University of Rochester AI Policy

Instructors should use course learning outcomes to determine when students must, may, or cannot use GenAI.

Instructors should create and communicate student GenAI course policies. For each assignment, the policy should state when students must, may, or cannot use GenAI and how they should verify, disclose, document, and attribute any GenAI use.

If an instructor has not explicitly stated that GenAI is allowed for an assignment, assume it isn’t.

Unauthorized GenAI use may be considered academic misconduct, similar to plagiarism or other unauthorized assistance.

Cheating: Don’t use GenAI to complete assignments unless you have explicit permission.

Usage of material generated by AI tools (Grammarly, ChatGPT, DALL-E, translation software, or similar) that is not allowed by the instructor.

Using unauthorized technology during an examination.

Instructors should use course learning outcomes to determine when students must, may, or cannot use GenAI.

All GenAI use must align with the applicable academic integrity standards [9] [10] [11].

Potential uses for students include using GenAI as a personal tutor, summarizing information, generating ideas, drafting documents, and self-testing for understanding [3] [4] [5] [6].

Use GenAI to Support Learning

GenAI can be a valuable tool, but remember that university studies are about building students’ own skills and knowledge.

Students should use GenAI to enhance—not replace—their learning.

GenAI can make mistakes or introduce biases. Students should always double-check GenAI-generated information for accuracy before using it in their work. For instance, if they ask GenAI to create a study guide, they should cross-reference its answers with their course materials or ask a teaching assistant or instructor to be sure it is accurate.

Instructors should create and communicate student GenAI course policies. For each assignment, the policy should state when students must, may, or cannot use GenAI and how they should verify, disclose, document, and attribute any GenAI use.

Usage of material generated by AI tools (Grammarly, ChatGPT, DALL-E, translation software, or similar) that is not allowed by the instructor.

Submission of work such as laboratory reports, computer programs or coding, journals, reflections, or other types of papers, which have been copied from work done by other students, either in whole or in part, with or without these students’ knowledge or consent.

Consistent with disciplinary and publishing norms, the use of GenAI should be disclosed/identified at the time of proposing, reporting, publishing or when engaging in other dissemination activities.

While the norms around authorship are evolving, if the final work contains more than de minimis AI-generated content as defined by your field, the nature and use of GenAI should be disclosed, and the parts of the work created by the author and by the GenAI should be distinguished and described.

Researchers have a duty to verify the output of GenAI systems to ensure that they have not introduced inaccuracies or biases into the research process.

Interpretation is an act of authorship: consider your discipline’s standards when GenAI is used to develop interpretations, and be aware that you are responsible for the integrity of the interpretation (e.g., how theory connects to the data, limitations for discussion, etc.).

Data Falsification: Don’t use GenAI to generate or alter data in ways that misrepresent your findings.

Three key duties of data protection, verification, and transparency frame appropriate use of GenAI in research.

Beyond the short list above, researchers may also be obligated to follow the policies of external entities regarding GenAI, including funding agencies, publishers, suppliers, or other institutions where collaborating researchers reside.

Consistent with disciplinary and publishing norms, the use of GenAI should be disclosed/identified at the time of proposing, reporting, publishing or when engaging in other dissemination activities. In reporting such use, the following existing policies should be adhered to:

More recently (December, 2023), NSF issued a very similar Policy Memo to its extramural research community, indicating that “NSF reviewers are prohibited from uploading any content from proposals, review information and related records to non-approved generative AI tools”.

Transparency: Faculty, staff, peer educators, and students should disclose when their work has been created, whole or in part, with a GenAI tool. Disclosures should specify how GenAI was used and, when appropriate, reflect on potential biases.

Instructors should create and communicate student GenAI course policies. For each assignment, the policy should state when students must, may, or cannot use GenAI and how they should verify, disclose, document, and attribute any GenAI use.

Credit to source material or source technologies (DALL-E, ChatGPT, or similar) must be given regardless of whether the idea, phrase or other material is quoted directly, or whether a student subsequently paraphrases or summarizes into their own words. In addition to any and all other citation information required (e.g., page numbers), verbatim quotes must always be clearly identified as such within the text.

Consistent with disciplinary and publishing norms, the use of GenAI should be disclosed/identified at the time of proposing, reporting, publishing or when engaging in other dissemination activities.

While the norms around authorship are evolving, if the final work contains more than de minimis AI-generated content as defined by your field, the nature and use of GenAI should be disclosed, and the parts of the work created by the author and by the GenAI should be distinguished and described.

GenAI detection tools are unreliable, biased, easily defeated, and unable to provide definitive evidence of academic honesty policy violations. If instructors use GenAI detection tools in a course, they should disclose to students when and how the software will be used. Instructors should avoid using GenAI detection software as the sole basis for an academic honesty policy violation; instead, they should use it to converse with the student and conduct further investigation as needed.

Unauthorized GenAI use may be considered academic misconduct, similar to plagiarism or other unauthorized assistance.

Usage of material generated by AI tools (Grammarly, ChatGPT, DALL-E, translation software, or similar) that is not allowed by the instructor.

When instructors use GenAI tools to develop teaching materials, they should disclose and document their GenAI use to model professionalism, transparency, and academic honesty within the classroom. Instructors are responsible and accountable for all materials created. As such, they should thoroughly review all materials for accuracy and bias and revise as necessary before use with students and course staff.

Instructors and course staff should maintain oversight of grading and feedback for student work. Grading and feedback are core teaching responsibilities that require human oversight.

Instructors should refrain from delegating core course responsibilities to GenAI.

If students are working as teaching assistants or in other peer educator roles, they should talk to the instructor about expectations for using GenAI. They should ask about GenAI’s role in their duties, such as grading, feedback, or instruction. Unless the instructor says otherwise, assume that using GenAI in these roles is not permitted.

Users should not upload to public GenAI platforms confidential and/or proprietary information, including moderate or high-risk data from the University of Rochester.

Non-public or sensitive University information should never be uploaded into public GenAI tools—whether free or paid—unless there is a university agreement with the vendor approved by one of the various AI governance groups.

I encourage you to take advantage of our secure version of a generative AI chatbot, designed to securely handle medium- and high-risk institutional data. Accessible at chat.rochester.edu, this advanced AI tool is exclusively for our faculty, staff, and students, providing a reliable and secure platform for managing a variety of inquiries and tasks.

Never enter…

Confidential, proprietary, or competitive information

PHI, PII, patient, or personal data

High Risk information must be protected even if the data are allowed to be shared outside the University. Disclosure of High Risk Information to a third party agent or vendor is permitted only if the agent or vendor assumes a legally binding obligation to safeguard the use and disclosure of the information (unless applicable law or regulation expressly allows for the disclosure without such a safeguard).

This document establishes guiding principles for the responsible use of GenAI in teaching and learning at our university.

As the technology matures, risks may evolve, and new mitigation strategies may emerge, underscoring the need for regular review and updates to ensure our guidelines remain relevant and practical.

Non-public or sensitive University information should never be uploaded into public GenAI tools—whether free or paid—unless there is a university agreement with the vendor approved by one of the various AI governance groups.

### AI Governance Council

Chair: Nicole Sampson, Provost

Co-Chair: Julie Myers, University CIO