The disclosure obligation runs the other direction
Most institutional AI policy in education has focused on one direction: telling students what they are and are not allowed to do with AI tools on assignments. The University of Illinois Chicago's new AI Use and Disclosure Statement, published by the Office of the Vice Provost for Faculty Affairs and reported on November 6, starts from the opposite end of that relationship. It requires instructors to disclose any planned AI use in their own teaching before the conversation about student use even begins.
The disclosure requirement is specific rather than general. It names course design, instructional materials, assessment, grading support, and feedback generation as activities instructors must disclose, covering both pre-class preparation work and activities that directly touch student work. That scope means a professor using AI to draft a rubric or generate feedback comments on an essay falls under the same disclosure obligation as a professor using AI to build a semester's syllabus from scratch.
A framework, not a single blanket rule
UIC's policy explicitly avoids setting one university-wide rule for what students can and cannot do with AI. Instead, each course syllabus must state which AI uses are acceptable and which are prohibited for that specific course's activities, assignments, and assessments, with instructors responsible for setting the conditions, including whether students themselves must disclose their own AI use. That puts the judgment call where the subject-matter expertise actually sits, with the instructor teaching the material, rather than with a university-wide committee trying to write one rule flexible enough to cover every discipline on campus.
That course-level framework acknowledges something most blanket AI bans and blanket AI permissions both ignore: a writing-intensive seminar and a quantitative methods course have entirely different legitimate uses for AI, and forcing them under one policy produces rules that fit neither well. Syllabi carrying these AI provisions must be distributed on the first day of class, with notice required for any substantial changes made later in the term, which gives the disclosure obligation actual teeth rather than leaving it as aspirational guidance buried in a faculty handbook nobody reads.
The accommodations clause closes a real gap
One line in the policy stands out for addressing a problem most AI education policy has not touched directly: AI tool expectations must not conflict with approved student accommodations. That single sentence prevents a scenario that was becoming increasingly plausible as AI restrictions tightened across higher education, where a blanket ban on AI tools for an assignment could inadvertently strip away a tool a student with an approved accommodation is legally entitled to use.
Getting this detail into a published policy, rather than leaving it to be sorted out case by case after a complaint, suggests UIC's faculty affairs office thought through the policy's edge cases before publication rather than after a dispute forced a revision. That is a meaningfully different development process than the reactive AI policy drafting most institutions, and most employers, have defaulted to over the past three years, where the accommodations question tends to surface only after a student or an employee files a formal grievance over a rule nobody had stress-tested in advance.
Why instructor-side disclosure is the harder policy to write
Writing a policy that restricts what students can do is, relatively speaking, the easier half of AI governance in education. Writing a policy that requires faculty to disclose their own use of the same tools is harder, because it asks the people with institutional authority and tenure protection to accept the same transparency obligations they are imposing on the people with the least power in the classroom relationship. It is the kind of policy that is simple to draft in principle and genuinely uncomfortable to enforce in practice, which is likely why so few institutions have published one.
UIC choosing to publish that requirement formally, through the Vice Provost for Faculty Affairs rather than as informal departmental guidance, signals institutional commitment to the idea that AI transparency obligations should not stop at the point where they become inconvenient for the people writing the rules. Whether individual faculty comply consistently is a separate and harder question the policy's existence cannot answer on its own, but publishing the expectation at all, in writing, with an office of record attached to it, is further than most peer institutions have been willing to go so far.
The enterprise parallel most companies are avoiding
Replace faculty with managers and students with employees, and UIC's policy maps almost directly onto a gap in most corporate AI governance frameworks right now. Plenty of companies have published AI usage policies for employees covering what tools they can use for which tasks. Far fewer have published a parallel policy requiring managers and leadership to disclose when AI influenced a performance review, a hiring decision, or a restructuring recommendation that affects the people reporting to them, even though those decisions carry far more legal and reputational exposure than a student's homework.
UIC's approach is a reasonable template for any enterprise technology or HR leader looking to close that same gap before it becomes a trust problem rather than a policy gap. A disclosure obligation that only runs downward, from management to staff or from institution to student, protects the institution's liability more than it protects the relationship. One that runs in both directions, with specific, auditable triggers rather than vague good-faith language, is a harder policy to write but a much more durable one to defend later.
