Carnegie Mellon University

The Eberly Center recommends extreme caution when attempting to detect whether student work has been aided or fully generated by AI. Although companies like Turnitin offer AI detection services , none have been established as accurate.

Collaboration or assistance on academic work to be graded is not permitted unless explicitly authorized by the course instructor(s). The citation of all sources is required.

Unauthorized assistance refers to the use of sources of support that have not been specifically authorized in this policy statement or by the course instructor(s) in the completion of academic work to be graded. Such sources of support may include but are not limited to advice or help provided by another individual, published or unpublished written sources, and electronic sources.

Additional guidelines and procedures for graduate students may exist at the college/department/program level, in which case they are communicated in the college/department/program graduate student handbook.

When assessing responsibility for an academic integrity violation, the course instructor should apply a preponderance of the evidence standard. A preponderance of the evidence means that the course instructor has determined that it is more likely than not that the student has violated the University Policy on Academic Integrity.

Suspected violations should be investigated within a reasonable timeframe, ideally within one to two weeks of discovering the alleged violation.

Instructors should notify the involved student(s) of the alleged violation(s) and provide the student(s) the opportunity to respond in an in person meeting or by phone or videoconference as circumstances warrant.

The most severe outcome that can be assessed at this level of review and action is course failure although the instructors and/or department heads may recommend additional outcomes for consideration during an Academic Review Meeting.

In extenuating circumstances, the course instructor can request that the student be withdrawn from the course in consultation with the Associate Dean for the college where the student is enrolled.

Students who have pending or documented academic integrity actions may not drop or withdraw from the course in question (including the use of a voucher).

Students who have pending or documented academic integrity actions may not change a course to P/NP (including the use of a voucher).

An academic integrity violation is considered pending once any version of the assignment has been submitted, regardless of when the violation is discovered or reported.

Students will receive official notification that the report has been received from the Office of Community Responsibility. The notification will include an invitation to meet with a process advisor in the Office of Community Responsibility to discuss the case, review the appeal process, and prepare for an Academic Review Meeting (if needed).

When a student has received official notification of course-level action from the Office of Community Responsibility, a student has seven calendar days to request an appeal of course-level action. All requests for appeal must be submitted in writing to the Office of Community Responsibility.

The written appeal request must indicate whether the student is appealing the finding of responsibility, the outcome, or both, as well as the basis for the appeal itself.

In the event of an appeal, outcomes will normally be held in abeyance pending the outcome of the appeal. In the event of an appeal at the end of a semester that affects recording of a course grade, an incomplete grade will be assigned pending the outcome of the appeal process.

Carefully review your course syllabi to understand what is and is not acceptable in each of your courses in terms of using generative AI tools in your coursework.

The university’s policy on academic integrity is not changing as a response to generative AI tools. According to CMU’s Office of Community Standards and Integrity, CMU’s academic integrity policy already implicitly covers such tools, as they may be considered a type of unauthorized assistance. However, this policy is intentionally designed to allow instructors to define what is “authorized” vs. “unauthorized” and what constitutes “plagiarism” and “cheating.”

A growing list of tools have been vetted by CMU that are FERPA compliant and therefore able to be used for teaching and learning purposes. GenAI tools currently licensed by Computing Services include: Microsoft Copilot, ChatGPT edu, Google’s Gemini, and NotebookLM.

Not all tools listed on Computing Services site are FERPA compliant and are typically indicated as such

The Eberly Center recommends extreme caution when attempting to detect whether student work has been aided or fully generated by AI. Although companies like Turnitin offer AI detection services, none have been established as accurate.

In addition to false positives and false negatives, detection tools may often produce inconclusive results. A detection tool can provide an estimate of how much of a submission has the characteristics of AI-generated content, but the instructor will need to use more than just that number to decide whether the student violated the academic integrity policy.

Even very strong evidence that a student used AI may be irrelevant unless you have a clear academic integrity policy establishing that the student’s use of the AI tool constitutes “unauthorized assistance” in your course.

Therefore, if you plan to use these tools in your course, you must provide an alternative for students who choose not to create an account.

The following examples represent a range of options one could adapt or adopt, based on their teaching context and course’s student learning objectives.

AI Gateway | AI Developer tool makes AI models easier to access. | Accessing management to models via API keys. | Faculty and Staff | No cost; however an Oracle String is required to pass on any costs incurred from using a model endpoint.

ChatGPT Edu | Synthesizes info, summarizes, and generates content. | Research, writing, and problem-solving. | Faculty and Staff; Students sponsored by a faculty or staff member | Tiered subscription model based on credit usage

Google Gemini | Google's AI that handles text, voice, and files. | Meeting summaries, working with Google Apps, and searching across formats. | Students, Faculty, Staff, and Sponsored | Available with your Andrew account.

Microsoft Copilot Chat | AI support for Office tasks, images, and code. | Microsoft files, documents, and coding help. | Students, Faculty, and Staff | Available with your Andrew account.

NotebookLM | Creates AI-powered notebooks from uploads. | Organizing ideas, summarizing sources, and creating podcasts. | Students, Faculty, Staff, and Sponsored | Available with your Andrew account.

Zoom AI Companion | Adds summaries and action items to meetings. | Real-time help during virtual meetings. | Students, Faculty, Staff, and Sponsored | Available with your Andrew account.

Faculty and staff Students (with faculty or staff sponsorship)

Small (Standard): 260 credits/month — $240 billed annually ($20/mo) Medium: 1,020 credits/month — $600 billed annually ($50/mo) Large: 2,300 credits/month — $1,200 billed annually ($100/mo) X-Large: 4,880 credits/month — $2,400 billed annually ($200/mo)

Pay-as-you-go model, with costs depending on the model and volume used. You only pay for the input and output tokens that you generate.

When you access one of CMU’s approved AI tools through a protected environment: You retain control over how your data is stored, processed, and reused. Your data remains private and secure. Your data will not be sold to advertisers or used to train AI models.

General Sharing: Custom GPTs, Google Gems, or Microsoft Copilot Agents that include Private, Restricted, or Restricted-Specified data as defined in CMU’s Data Classification Guidelines may only be shared with individuals who have the appropriate access rights, as determined by the CMU Data Stewards.

All active CMU faculty, staff, and students with a licensed Zoom account already have access to Zoom AI Companion when they log in with their Andrew accounts.

IMPORTANT: When you log in with your CMU account, Zoom does not use your prompts and responses for AI training, and uploaded documents are not retained beyond your session.

When a second (or higher) violation is reported to the Office of Community Responsibility (OCR) or there is a request for additional action and the violation is not being appealed, a meeting will be scheduled

The typical outcome(s) for a student with two violations of the University Policy on Academic Integrity includes a one or two-semester suspension and restorative outcomes. Additional and/or more severe violations may increase the length of separation, including the possibility of permanent separation (expulsion).

Students participating in an Academic Review Meeting may have a Carnegie Mellon University community member (faculty, staff, student) serve as a support person. A support person will not be permitted to participate in the meeting but may advise the student and/or be present during the meeting.

Students who choose to have a non-affiliate as their support person will have their case reviewed by the Academic Responsibility Panel.

The student, the course instructor, and/or the student’s department head or designee may submit a request for appeal of the Academic Review Meeting decision in writing to the provost within seven (7) days of being notified of the outcome(s) under the following circumstances:

The outcomes determined during the Academic Review Meeting were not commensurate with the violation(s) of the University Policy on Academic Integrity for which the student is responsible; New information, which was not reasonably available at the time of the review, merits re-consideration of the final disposition.

Each case presented to the Academic Responsibility Panel is heard by two student panel members and three faculty panel members.

It is the responsibility of the Office of Community Responsibility to ensure that times set for panels are reasonable and that necessary accommodations have been offered to both the student and course instructor(s) whenever appropriate.

The student may submit a written introduction and up to three written letters of support by a date specified by the process advisor.

A student may appeal a decision delivered by the vice provost for education or designee following the Academic Responsibility Panel process by submitting a request for appeal in writing via email to the provost of the university within seven calendar days of the official notification of the decision.

In the event of an appeal, outcomes will normally be held in abeyance pending the outcome of the appeal. In the event of an appeal at the end of a semester that affects recording of a course grade, an incomplete grade will be assigned pending the outcome of the appeal process.

Appeals will be decided by the appeal officer in a timely manner as circumstances warrant. While an appeal is under review, the appeal officer will update the student, course instructor, and department head as necessary about the anticipated timeline.

The academic integrity actions outcomes defined below comprise a range of official actions that may be imposed during an Academic Review Meeting or Academic Responsibility Panel. One or more of these actions may be imposed in response to a given situation. This list is not comprehensive, as other outcomes may be imposed as appropriate.

Drop from Academic Program: The student is required to sever entirely their connection with the academic program in which they are currently enrolled. Students who have been dropped from an academic program may apply for admission to other academic programs at the university.

Expulsion: The student is required to sever entirely their connection with the university. Students who have been expelled are required to absent themselves from university property (including but not limited to university housing, offices, labs, and libraries) within a maximum of two days after the action and to remain off the campus permanently.

The determination of outcomes will be guided by the following considerations: the interests of the community, the learning and development of the student(s), the impact of the violation(s), and any mitigating or aggravating circumstances.

AI works best as a starting point, a companion for reflection, or a tool to organize your thinking. It should never replace your own judgment, authentic voice, or decision-making.

When accessing any CMU‑provided AI platform, you must sign in with your Andrew Account credentials and complete CMU‑branded Single Sign‑On to ensure the correct level of security, service, and university‑approved support.

Avoid using third-party apps to automate application submissions, as employers flag or filter AI-autonomous submissions.

Never use live AI tools during an actual interview. Employers value genuine communication and natural storytelling.

GenAI can help you overcome writer’s block, compare phrasing, and tailor materials. However, the CPDC strongly discourages using chatbots to generate the majority or entirety of your resume bullet points.

Many library electronic resources are governed by license agreements that may restrict uploading articles, ebooks, PDFs, datasets, or other content into third-party AI platforms.

Do not upload sensitive, confidential, unpublished, or copyrighted materials into public AI tools unless you have permission to do so.

All institutional data should be classified into one of four sensitivity levels or classifications:

Restricted-Specified | Data that is classified as restricted but also has additional requirements for protection based on sponsors, contracts, regulations, and/or data use agreements.

Just as it is critical to include a syllabus policy for student use of genAI, it is also important to explicitly state how instructors and TAs will use genAI (if at all).

An "Academic Action" is an action by a program, unit or department based on a graduate student's academic performance or failure to satisfy academic program requirements.

Graduate students who wish to appeal an Academic Action must submit a formal, written appeal, with appropriate documentation, to the Dean of the College within seven (7) calendar days after receipt of written notice of the academic action by the department, unit, or program head or director, or designated committee.

Graduate students who wish to appeal the decision of the Dean must submit a formal written appeal, with appropriate documentation, to the Provost within fourteen (14) calendar days after receipt of the Dean's written decision.

We recommend that you adopt an academic integrity policy that considers the following: Whether or not AI tools are considered authorized or unauthorized assistance and in what circumstances.

We recommend that you adopt an academic integrity policy that considers the following: Whether or not AI tools are considered authorized or unauthorized assistance and in what circumstances. How students should cite assistance from or content/ideas generated by either AI tools or humans.

Consumer tools that have not been vetted by the university and/or are not FERPA-compliant (i.e., student work, which is part of their academic record, is being shared with third-party individuals and/or platforms) means that instructors cannot legally require students to create accounts and/or use these tools to complete course assignments.

Some instructors may be interested in using an AI tool to grade student work. If you choose to do so, every effort must be made to anonymize the student’s work by not connecting that work to any Directory Information on the student

Check over AI-generated outputs and incorporate human edits/revisions. Do NOT assume that quality grading occurred or haphazardly share unreviewed genAI feedback with students.

They may be used (including Custom GPTs, Google Gems, and Microsoft Copilot Agents) with any CMU data except Controlled Unclassified Information (CUI). Always follow CMU’s Data Classification Guidelines and any sponsor/regulatory/contractual requirements.

The AI tools listed on this page are available at CMU. When you sign in with your Andrew ID and password, each tool is FERPA-compliant and will not use your data to train its AI models.

AI Gateway access is team-based. Create multiple teams to control access, track spending, and process invoices.