NYU Tandon releases traffic-light AI policy guide for faculty
TL;DR
- NYU Tandon's Protecting Learning guide, released November 12, 2026, pushes faculty to classify assignments by whether AI advances or replaces learning.
- A traffic-light framework replaces blanket bans; one foundational course showed near-100% homework scores against 50% proctored exam averages.
- 80% of full-time Tandon faculty joined the review, 13 departments filed written plans, and 225 students voted on policies; public comment runs through November.
One foundational NYU Tandon course logged homework scores near 100% while proctored exams averaged 50%, a gap the school cites as the kind of thing its new AI-use playbook is built to catch. The 18-author guide, Protecting Learning: A Practical Guide to AI-Use Policy for Instructors, Departments, and Schools, was published on November 12, 2026 by a faculty working group inside Tandon's AI Across the Curriculum initiative, with public comment open through November and a first revision planned for January 2027.
The framework rejects both blanket bans and open use. Assignments get sorted into a traffic-light system: red where AI is prohibited and students must show independent competency, yellow where AI is permitted in limited supporting roles, and green where effective AI use is itself part of the skill being taught.
The school's own rollout gives the document some weight. According to NYU Tandon, 80% of full-time faculty participated in a school-wide review, 13 departments submitted written plans, and 225 students across two Responsible AI courses voted on policies. Leading the effort was Julia Stoyanovich, Associate Dean of AI Initiatives and director of Tandon's Center for Responsible AI. 'Guidance alone does not change classrooms; ownership does,' she said. Learning loss, defined as students submitting correct work without mastering the material, is named as the leading concern, ahead of cheating.
The guide is not an NYU mandate and sets no single policy for Tandon courses. Its authors describe it as adaptable to other departments and institutions, built on the principle that 'effective AI-use policies should not only establish clear expectations but also be developed through a process that engages the people they affect.' Two of the researchers we track posted the link within a day of its release.
Shared on Bluesky by 2 AI experts
-
Thoughtful guide from @nyutandon.bsky.social professors on course design in the era of AI ultraviolet.library.nyu.edu/records/xp9w... Comments welcome: tandonai.netlify.app
View on Bluesky →
Originally reported by ultraviolet.library.nyu.edu
Read the original article →Original headline: Protecting Learning: A Practical Guide to AI-Use Policy for Instructors, Departments, and Schools