AI and Academic Integrity

This website offers SDSU faculty a summary of the academic integrity issues related to generative AI. The resources below reference recent Senate-approved policies, and assemble some pedagogical recommendations relevant to faculty considering AI and academic integrity issues.

While this website doesn’t (and cannot) offer a comprehensive solution for every AI-related academic integrity issue that may arise, CTL has developed a set of topics and resources to guide faculty in the necessary conversations ahead. We recognize that specific disciplines, students, and courses will have specific demands. How you respond to the AI-related academic integrity issues that may surface in your course is up to your judgement as the instructor.

We encourage faculty to recognize that many SDSU students feel pressure from multiple directions. Behaviors that get called cheating have “a host of causes that go well beyond dishonesty” (Bertram Gallant and Rettinger, p. 4). The CTL invites faculty to “reframe cheating as an opportunity to educate” when responding to student misuse of generative AI (p. 225).

Additionally, your department, school, or field may have its own policies about generative AI. Check with your chair or director and consult your colleagues for resources in your field. Faculty are encouraged to have regular conversations about what best practices for AI and academic integrity look like in your discipline (especially around issues like AI and grading; see below) and what you collectively feel will best support the integrity of your degrees and serve your students.

The University Senate has recently approved several policies related to generative AI that have implications for your classes.

The SDSU Senate Policy File states that all syllabi must now include a “statement of permitted use of generative AI, which may vary by section.” (Senate Policy File, p. 152 [PDF p. 158], under University Policies: Faculty, Academic Responsibilities, §2.0 Course Syllabi)

An AI syllabus statement gives your students clear expectations for Generative AI use. Beyond that, it is a critical starting place for a longer (ongoing) conversation with your students about the cognitive capacities the course is designed to support and how genAI use intersects with your learning outcomes and their education more generally.

SDSU’s Syllabus Template has guidance related to AI Syllabus Statements (pp. 4-5), including sample text for you to use or adapt to suit your course.

The SDSU Student Code of Conduct now defines plagiarism to include “representing work produced by generative artificial intelligence (AI) as one’s own.” See “Cheating and Plagiarism.” This language points students to the importance of doing the core “work” of the course and developing essential cognitive capacities. Faculty can refer to this language as a touchpoint in ongoing conversations with students.

The CTL recommends that faculty take “approaches to academic integrity that support students rather than punish them and that promote a collaborative rather than adversarial relationship between teachers and students” (MLA-CCCC p. 10).

The Senate Academic Responsibilities policy now requires instructors to “disclose to their students any use of generative AI in the creation of instructional materials” and to explain their use of AI in order to “build trust and model metacognition, intentionality, and transparency for their students.” (Senate Policy File, p. 151 [PDF p. 157], §1.3.)

Instructors of record shall. Instructors should provide a rationale when disclosing their own genAI use in order to build trust and model metacognition, intentionality, and transparency for their students.

Just as you may ask students to be transparent about their AI use, it is worth being transparent about your own, whether you are using a GenAI app for designing slides, generating practice problems / discussion questions, or assisting with feedback on student work. A short disclosure statement in your syllabus and / or on relevant assignments does two things: it models the transparency that you are asking of students, and it protects you if questions ever arise about how any course material was produced or a grade was determined.

A useful starting template, adapted from a framework recently developed by the Modern Language Association’s AI and Research Working Group, might read something like:

This course/assignment used [tool, version] for [purpose] at [stage (e.g., drafting practice questions, generating a rubric)]. All content was reviewed and revised by the instructor.

Adapt the bracketed portions to your own use.

The use of GenAI tools in grading requires consideration. The same Senate policy states that in the interest of transparency faculty “shall disclose to their students any use of generative AI tools in grading.” The policy adds: “Students should also be given the opportunity to dispute grades when AI is utilized as a measure of determining grades on assignments.” (Senate Policy File, p. 70 [PDF p. 76], §3.2.1.)

The CTL recommends that courses using GenAI tools in the review of student work should keep a human in the loop: an instructor or grading assistant should check student work or otherwise monitor the process before grades are finalized.

The point isn’t a fixed formula for faculty disclosure of AI use but a practice of acknowledging the role of a specific tool in the creation of instructional material or the review of student work. The bottom line is making it clear to your students that they are taking the class from you, not an AI app.

Faculty may find it useful to incorporate AI-use reflections as part of take-home assignments. By including a reflection as part of an assignment, faculty assert “consistent, equitable expectations for transparency that require students to document and disclose their use of any resources, including AI tools” (Frontier 2025). Asking students to disclose if and how they may have used GenAI on an assignment “gives students the opportunity to think critically and openly about their AI usage” while giving faculty the opportunity to communicate in advance what they think is appropriate, ethical use or nonuse of AI tools for a given assignment (Watkins 2024). A faculty member can model such AI disclosure in a syllabus statement summarizing their own use of GenAI tools for the creation of instructional materials. Incorporating AI reflections in assignments will help normalize disclosure of AI use (Watkins 2024).

Despite their promise, applications designed to detect writing produced by generative AI are unreliable.

Here at SDSU, the AI content detection feature of Turnitin.com is not activated. Other CSU campuses have this feature disabled, too; so do many other colleges and universities across the country. The reason is that Turnitin’s AI detector is prone to errors, as are similar applications from other providers.

Faculty turning to AI detectors are advised to use caution. False positives and false negatives are not unusual; for examples, see this informal (at times playful) essay about AI detectors by SDSU colleague Jacob Hubbard. Further, researchers at Stanford demonstrated that AI detectors are consistently biased against students working in a second language, a bias that could impact many SDSU students.

Though CTL does not recommend use of AI detectors, faculty who still intend to use an AI detector should mention that in your syllabus (along with any other required technology). The full ramifications of the use of this technology for student privacy are still not known.

Finally, faculty should take the results of a prompt in an AI detector as the starting point for a discussion with a student rather than as definitive proof one way or another.

Course modality, size, format, discipline, and level of instruction (lower division, upper division, graduate seminar) may shape how faculty address GenAI and academic integrity. Here are some resources for thinking it through and discussing with colleagues:

Large enrollment / asynchronous courses. Some experts argue that generative AI exacerbates the challenges of teaching in large, low-contact formats. As you develop your own approach, this podcast episode from experienced CSU colleagues may be helpful: Are Asynchronous Online Classes Broken?

Surveillance technology is seen by some faculty as a useful tool for testing in asynchronous (fully online) courses. Several units on campus are piloting testing centers for students who can come to campus. Lockdown browsers have been in use for years, growing in popularity during the pandemic; they are seen by some as a way to mitigate student misuse of GenAI during online testing. The topline pros and cons of these tech tools are mitigating academic misconduct versus student privacy concerns.

It is important that faculty have conversations about these tools with faculty in their units. Instructors who choose to use surveillance technology in their classes should make sure to clearly state that intention in their syllabus and regularly communicate with students about it.

Writing Across the Curriculum. The CTL is convening faculty experts to develop recommendations for faculty teaching writing across the curriculum. In the meantime, faculty with writing assignments in their courses are encouraged to consult this background paper on Writing and AI, jointly prepared by the Modern Language Association and the Conference on College Composition and Communication.