NYU Tandon Unveils a New Strategy and Playbook for AI in the Classroom

NYU Tandon School of Engineering released a framework for governing generative AI in the classroom that rejects both blanket bans and unrestricted use. It asks instructors to decide, assignment by assignment, whether AI advances learning or replaces the work students need to do themselves.

The framework is laid out in Protecting Learning: A Practical Guide to AI-Use Policy for Instructors, Departments, and Schools, published by NYU Tandon this week. The guide is open for public comment, with feedback received during November informing its first revision in January 2027.

NYU Tandon also plans to publish a summary of the comments received and the changes made in response.

Julia Stoyanovich, NYU Tandon's inaugural Associate Dean of AI Initiatives, led the effort, which brings faculty and students into decisions about how AI should be used in teaching and learning. The guide is part of Tandon's broader AI strategy, coordinated centrally by Stoyanovich and tailored to individual disciplines, that includes offering new AI courses to both students and faculty.

Protecting Learning arrives as universities debate how to embrace AI without letting it undermine learning. An MIT committee warned this summer that AI can erode learning when it does students' intellectual work for them, even as it can be a powerful tool when it supports their thinking. Institutions remain split on where to draw that line, with some building AI into curricula and others restricting it and adding in-person assessment.

“Guidance alone does not change classrooms; ownership does,” said Stoyanovich, a professor of computer science and engineering and director of Tandon's Center for Responsible AI. “Our students surprised me by choosing stricter AI rules than I would have set. Students learn by working through problems themselves, not simply accepting answers from a machine. The same is true of institutions. Bringing faculty and students together to work through these questions is just as important as the policies they produce. We hope other institutions will adapt this guide, challenge it, and tell us what they learn.”

The guide recommends instructors sort their assignments into three categories. Red, where AI is prohibited because students must show they can do the work independently; Yellow, where AI can play a limited supporting role; and Green, where using AI effectively is part of the skill being taught. Most courses will use more than one.

The framework emerged from a school-wide review. Over the spring and summer of 2026, every department examined how AI was affecting its courses. About 80 percent of full-time faculty took part, and 13 departments and programs submitted written plans. Faculty working groups then compared those plans with emerging research and practices at other universities. The guide's co-authors include faculty from each Tandon academic department and four undergraduate and graduate students.

The biggest concern was not cheating but learning loss, or students turning in correct work without mastering the material. In one foundational course, homework scores approached 100 percent while proctored exams on the same material averaged about 50 percent.

The guide therefore recommends AI-free checks, such as in-class exams, oral questions, or proctored quizzes, wherever students need to demonstrate independent mastery.

Students also helped set the rules. This fall, about 225 students in two Responsible AI courses debated and voted on their own AI policies, from AI-generated code to AI use in writing and readings. Majorities in both courses rejected unrestricted AI use for written work, and differences between the two groups strengthened the case against one-size-fits-all rules.

The guide also cautions against relying on AI-detection software. It recommends treating suspected unauthorized AI use as the start of an inquiry, not a finding of misconduct, beginning with a conversation with the student and weighing multiple kinds of evidence. Detector scores alone should not establish a violation, the guide says, citing false positives and higher risks for multilingual and non-native English writers.

The guide is not an official NYU mandate and does not prescribe a single policy for Tandon courses; instead, it is meant to serve as an open-source course policy model of sorts. It offers shared language, adaptable policy tools, student-engagement practices, and an academic-integrity process that instructors and departments at Tandon and other NYU schools—and outside institutions—can adapt. Tandon will track how these approaches work in practice.