The University of North Carolina system sent Education Secretary Linda McMahon a 10-page letter this week defending academic rigor in the age of artificial intelligence. The letter urged universities to verify that students have actually mastered the skills their degrees represent and called for clear classroom rules on AI disclosure.
But the letter itself was partially drafted with AI.
UNC system spokesperson Andy Wallace confirmed administrators used AI to compile existing materials - policies, regulations, speeches and other documents - for portions of the response. He called the use "typical of administrative tasks," noting that AI is increasingly used in business operations to synthesize large amounts of information.
The Chronicle ran the letter through three AI detectors - Pangram, GPTZero, and Winston - and all three flagged portions as highly likely to be AI-generated. Among the flagged sections was the very passage on AI expectations, which stated: "Human judgment must remain responsible for consequential academic decisions."
Detectors can't determine how much assistance administrators received, and their reliability is contested. But the episode highlights an awkward tension: students and professors face scrutiny for AI use, while administrators often don't.
Administrators face growing scrutiny over AI use
The letter arrives as academics face heightened attention for their own AI use. The Daily Californian reported this week that a UC Berkeley math professor who wrote a viral column about unprepared students admitted to using AI to help edit it.
Some administrators have been open about their AI practices. Texas Tech University used AI to review more than 14,000 courses for left-leaning content. Florida's Department of Government Efficiency used large language models to review 75,000 syllabi across the state's public universities, with roughly 15 percent flagged for not being "ideologically neutral or balanced."
UNC didn't initially disclose AI use in the letter, which was attributed to President Peter Hans and Board of Governors Chair Kellie Hunt Blue. Wallace disputed any hypocrisy in using AI for a document that asks students to disclose their own AI use, saying the letter's disclosure guidance applies to academic work, not administrative functions.
No systemwide AI policies - yet
UNC's website notes there are no systemwide AI policies. Individual campuses have published their own standards. Chapel Hill's online hub offers guidance on using AI "responsibly" and suggests staff "should acknowledge the use of generative AI in administrative tasks."
"Documenting how generative AI is sourced and why helps stakeholders understand the role of generative AI technologies in their interactions and decisions," the Chapel Hill website says. "Sourcing documentation is crucial to ensuring sustainability and the retention of institutional knowledge."
Lee Rainie, director of Elon University's Imagining the Digital Future Center, described the current landscape as "the Wild West." At the individual level, he said, "they're doing things to their own personal tastes and needs and moral judgments." Rainie drew a distinction between using AI for menial administrative tasks and using it for scholarship or coursework, where it risks compromising students' learning.
Administrators face calls for consistency
Cate Denial, a history professor at University of Minnesota Rochester who writes about AI in higher education, said administrators should be held to the same standards they set for faculty and students. She called the UNC letter hypocritical.
"Communications coming from the highest level need to be modeling the kinds of things that they're expecting of everybody else," she said. "If you're telling your students that faculty will disclose where they have used AI in the preparation of course materials, give the same directive to students, then say that 'We are exempt from those kinds of considerations,' it seems weak sauce."
Shortcuts are risky, Denial added, "because you end up breaking trust with the people you are supposed to represent."
"When we turn away from doing the work of communicating and representing other human beings," she said, "then I think we've lost our way."
Why this matters for education professionals
For faculty and administrators, the UNC episode is a practical reminder that AI policies rarely apply uniformly across an institution. If your campus expects students to disclose AI use, check whether the same expectation applies to staff communications and course materials. Establishing those expectations openly - before a news outlet runs the text through a detector - is easier than defending the inconsistency after the fact.
If you're an educator or administrator looking for practical guidance on AI policies and use cases, the AI for Education resource hub covers campus-specific applications. For those handling administrative writing tasks, the AI Learning Path for Administrative Assistants offers training on drafting documents efficiently while maintaining transparency.
Your membership also unlocks:
