Middlebury College AI Policy

Some of your faculty may invite you to use generative AI tools in your classes (often with restrictions or limits) and some faculty may restrict their use entirely.

Your faculty should include a syllabus statement indicating what is allowed and what is not allowed. As with other unauthorized aid, failure to comply with class policies prohibiting generative AI use could be construed as an Honor Code violation, which could result in disciplinary action. If a professor has not specified their policy on generative AI use, you should assume that it is forbidden unless the professor grants you permission verbally or in writing.

You are responsible for the accuracy and integrity of any work you submit, including any work that was created with support of an AI tool.

Some of your faculty may invite you to use generative AI tools in your classes (often with restrictions or limits) and some faculty may restrict their use entirely. Most of your faculty will have included a syllabus statement indicating what is allowed and what is not allowed. As with other unauthorized aid, failure to comply with class policies on generative AI use could be construed as an Honor Code violation, which could result in disciplinary action.

If a professor has not specified their policy on generative AI use, you should assume that it is forbidden unless the professor grants you permission verbally or in writing. You are responsible for the accuracy and integrity of any work you submit, including any work that was created with the support of an AI tool.

Honor Code Pledge: The Honor Code pledge reads as follows: "I have neither given nor received unauthorized aid on this assignment." It is the responsibility of the student to write out in full, adhere to, and sign the Honor Code pledge on all examinations, research papers, and laboratory reports. Faculty members reserve the right to require the signed Honor Code pledge on other kinds of academic work.

As with other unauthorized aid, failure to comply with class policies prohibiting generative AI use could be construed as an Honor Code violation, which could result in disciplinary action. If a professor has not specified their policy on generative AI use, you should assume that it is forbidden unless the professor grants you permission verbally or in writing.

As a student at Middlebury, you have access to several generative AI tools with your Middlebury credentials.

Middlebury is committed to helping our students engage critically with generative AI tools and to learn about their potential benefits and harms. We encourage you to participate in events and other learning opportunities that will deepen your understanding of generative AI tools.

In addition to learning about generative AI and AI in some of your courses, you may participate in related workshops and student-sponsored events. Training modules on generative AI are also available on LinkedIn Learning (log in with your Middlebury credentials for access). If you are looking for guidance from fellow students, DLINQ’s Interns are available to support you as you explore the potential and appropriate use of generative AI technologies.

Some of your faculty may invite you to use generative AI tools in your classes (often with restrictions or limits) and some faculty may restrict their use entirely. If a professor has not specified their policy on generative AI use, you should assume that it is forbidden unless the professor grants you permission verbally or in writing. You are responsible for the accuracy and integrity of any work you submit, including any work that was created with support of an AI tool.

Yes. There are concerns about the data that forms the foundation of the AI model and there are concerns about the privacy of your interactions with the model as well as any information you share. Some of these concerns may be addressed in the terms of service. We strongly discourage you from putting any personal information in generative AI models.

Proprietary research data and manuscripts, before public release.

Storage: Must be stored on secure, Middlebury-approved systems with appropriate access controls (e.g., password protection, multi-factor authentication). Cloud storage must be approved and configured for security.

Sharing: Sharing is restricted to authorized internal parties. External sharing requires explicit approval and appropriate data sharing agreements.

Public data is information that is intended for public consumption or is readily available to the general public. Its unauthorized disclosure would have little or no adverse impact on Middlebury, its operations, or individuals.

* Publicly available research data (after appropriate review and release)

* Storage: Can be stored on any Middlebury-approved system or publicly accessible platforms.

Sensitive data is information that is not intended for public release, and whose unauthorized disclosure, alteration, or destruction could have a moderate adverse impact on Middlebury, its operations, or individuals. This data typically requires protection to ensure confidentiality, integrity, and availability.

* Proprietary research data and manuscripts, before public release.

* Storage: Must be stored on secure, Middlebury-approved systems with appropriate access controls (e.g., password protection, multi-factor authentication). Cloud storage must be approved and configured for security.

* Sharing: Sharing is restricted to authorized internal parties. External sharing requires explicit approval and appropriate data sharing agreements.

Regulated data is information subject to specific legal, regulatory, or contractual obligations, whose unauthorized disclosure, alteration, or destruction would result in adverse impact, including significant financial penalties, legal liabilities, reputational damage, and harm to individuals. This data requires the highest level of protection.

Please reach out to Middlebury Information Security with any questions related to regulated data. Any use or storage of regulated data must be documented by the data steward. It must include a plan for data lifecycle and retention and must be registered with the Middlebury Information Security Team.

If your answer to all of the following questions is “yes,” then you must apply for IRB approval of your research.

Is IRB approval needed for classroom-based research activities?

It depends. Many courses involve student activities that would be considered human subjects research if they were occurring outside of the classroom context. The IRB has developed guidance to help you determine if you class-based research activities require IRB approval: Guidance for Classroom Research Activities.

How and when you cite AI will depend on the citation style you are following and any additional instructions given by your instructor. The APA has specific guidelines for citing AI as does the MLA. If your instructor has not provided guidance about how to cite AI-supported work, ask them how they would prefer for that work to be cited.

As with other unauthorized aid, failure to comply with class policies prohibiting generative AI use could be construed as an Honor Code violation, which could result in disciplinary action.

The policy expressly forbids the following acts:

* Plagiarism

* Cheating

* Duplicate Use of Work

* Falsifying Data

Therefore, in addition to adhering to the Academic Honesty Policy themselves, the student-written Constitution of the Undergraduate Honor Code states that “Any member of the College community (student, faculty, or administrator) who is aware of a case of academic dishonesty is morally obligated to report it to the professor or the Dean of Students."

AI should augment, not replace, the vital human relationships between teachers and learners. Faculty should maintain a primary role in both formative and summative evaluation of students’ learning progress, behaviors, and outcomes.

We provide support and resources for faculty to incorporate generative AI (and other technologies) into their courses. Faculty can also exercise considerable autonomy about whether and how to integrate generative AI into their courses and their work if they choose not to for pedagogical reasons.

As a Middlebury staff member, you may encounter generative AI tools like Microsoft Copilot, Google Gemini, Adobe Firefly, or Zoom AI Companion in your daily work. These tools, when used correctly, can help generate ideas, automatically take notes in meetings, and increase productivity. However, it’s essential to use them responsibly and within Middlebury’s policies.

Staff using AI tools are expected to adhere to Middlebury’s Responsible Use and Information Security policies. Be mindful of how AI can impact data privacy, security, and ethical standards. Some important practices to keep in mind:

* Avoid sharing sensitive information: Never input personal or confidential data (e.g., passwords, account credentials) into any AI tool.

* Stay engaged: Don’t rely on AI to operate autonomously—human oversight is always needed to guide and review output.

ITS is responsible for reviewing AI tools and providing information to users about what tools are available and how to get support for them. Visit ITS’ webpage on AI/generative AI for a current list of AI tools and additional information on the review processes for AI at Middlebury.

Faculty can use AI tools outside those made available by ITS. In doing so, you need to make sure that you understand the tool and any consequences of using it. Using non-sanctioned tools may involve risks to data privacy, security, and compliance with institutional policies.

* Avoid sharing sensitive information: Never input personal or confidential data (e.g., passwords, account credentials) into any AI tool.

* Storage: Must be stored on secure, Middlebury-approved systems with appropriate access controls (e.g., password protection, multi-factor authentication). Cloud storage must be approved and configured for security.

* Sharing: Sharing is restricted to authorized internal parties.

Please reach out to Middlebury Information Security with any questions related to regulated data. Any use or storage of regulated data must be documented by the data steward. It must include a plan for data lifecycle and retention and must be registered with the Middlebury Information Security Team.

While these shifts and their consequences are uncertain, we believe that liberal arts and sciences are key to mitigating negative outcomes and enabling positive outcomes. This moment presents an opportunity to reimagine the future of liberal arts and sciences and to create a distinctively Middlebury approach to comprehending, harnessing, and humanizing these technologies.

Faculty empowerment is essential to AI’s effective and ethical use in a liberal arts and sciences environment.

Middlebury will regularly consider and update what technologies are part of the “full Middlebury experience” and thus are made available to students, faculty, and staff through centralized facilities or individualized grants. Access includes availability of technologies and instruction in foundational skills and knowledge about those technologies, preparing learners to use AI proficiently, safely, and ethically.

We will establish effective policies, guidelines, and practices that center the accuracy, fairness and equality, accountability, and transparency of software programs and algorithms.

AI systems should never compromise the privacy of students’ personal and educational data. Automated decision making systems should only support human decision making, not replace it. Decisions made by AI systems should be transparent and able to be evaluated.

Learning involves exploration and risk-taking. Sometimes we will misstep or fail, and the fear of such mistakes should not impede forward movement. We will aim to learn from our steps and missteps by tracking and measuring outcomes, and to share all outcomes with openness and transparency.