ChatGPT policies at schools
How K-12 districts, universities, and major academic institutions are handling ChatGPT in 2026: who is banning it, who permits it with disclosure, and what students need to know.The arc so far
The first wave (2023) was bans. New York City public schools blocked ChatGPT on school networks, and several flagship universities issued tentative no-AI statements for coursework. The assumption was that AI output was categorically prohibited.The second wave (2024) was reversal. NYC public schools reversed the ban in May 2023. University faculties began drafting nuanced policies: allowed for brainstorming, not for drafting; allowed with disclosure; prohibited for specific assignments. Most policies moved toward case-by-case.The current state (2025 and 2026) is disclosure plus instructor discretion. Most major institutions now require students to disclose AI use, with specifics set by each course or assignment. A minority still prohibit AI outright in specific contexts, such as timed exams and first-year writing courses.University-level policies
Major universities have published AI policies that are broadly similar in shape:- Princeton's AI-in-the-classroom guidelines: instructors set course-level policy, and disclosure is strongly encouraged.
- Harvard FAS AI guidance: initial guidelines emphasize disclosure and instructor discretion.
- MIT's AI and teaching resources: faculty-facing materials on integrating or restricting AI per course goal.
K-12 policies
K-12 is more restrictive on average. Most school districts block public AI tools on their networks and devices, though many permit district-approved AI products, such as Khan Academy's Khanmigo and Google's education AI offerings, that are designed with classroom safeguards. Policies vary enormously by district and state.For K-12 students, the safe default is: do not use public AI tools for schoolwork unless your teacher has explicitly said it is okay. If in doubt, ask.LMS-integrated detection
Most universities and many K-12 districts use one of three LMS platforms: Canvas, Blackboard, or Moodle. Turnitin's AI writing indicator is integrated with all three and runs automatically on submitted papers. Students do not see the indicator; instructors do. For the current methodology, see the official Turnitin AI writing product page.Other detectors, including GPTZero's education products and Copyleaks, are also seeing adoption, but Turnitin is dominant because it is already embedded in academic workflows. See our guide to what AI detector Turnitin uses for how the integration works.What instructors are told to do
Guidance to faculty has converged on a pattern: use the detector score as one signal, compare against the student's prior work, and have a private conversation before formal charges. Most institutions no longer recommend initiating academic dishonesty proceedings based on a detector score alone.The MIT Communication Lab AI in writing guide is a widely shared example of how this conversation is framed.What students need to know
- Read the syllabus. The policy that matters is usually the course policy, not the university policy.
- Ask before assuming. If the syllabus is ambiguous, email the instructor. Written confirmation protects you.
- Disclose when in doubt. A short acknowledgment, such as "I used ChatGPT for brainstorming and the first draft; I revised substantially," is almost always better received than silence.
- Keep drafts. If you are accused of AI use, your revision history in Google Docs, Word, or git is the single strongest defense. Detectors produce signals; drafts produce proof.