Article
Cheating has become one of higher education’s most prominent worries, prompting many instructors to move assignments and exams back to paper. But that “blue book” retreat is a poor fit for programs that run fully online or lean on modern coursework tools, leaving many educators stuck in what a recent New York Times article described as the “obvious cheating, no obvious solutions” paradox.
By Megan Allen, Ed.M.
Partner Success Manager
Eight Areas
A recent study by Crompton et al. (2026) gathered expert perspectives from 22 countries to work through the fragmented policy debate on AI, creating guidance across eight areas:
01
Academic integrity
Define unacceptable substitution of AI for student work, clarify attribution and disclosure expectations, and align rules with assessment design.
02
Ethical and responsible use
Address accuracy, bias, transparency, accountability, and the conditions under which AI use supports rather than undermines learning.
03
Privacy and protection
Govern the entry of personal, confidential, research, and institutional data into external AI systems.
04
Equitable access
Prevent policy from assuming that all students and staff have equal access to capable tools, connectivity, or paid services.
05
Gen AI literacy
Develop the ability to use, question, verify, disclose, and critically evaluate AI outputs.
06
Integration strategy
Move beyond ad hoc tool use toward institutionally coherent decisions about where AI belongs in teaching, learning, research, and administration.
07
Human oversight and accountability
Ensure that people remain responsible for consequential decisions and that AI augments rather than replaces professional and academic judgment.
08
Institutional support and infrastructure
Provide training, approved tools, technical support, governance capacity, and other conditions needed to make policy workable.
Each of these areas maps onto a pain point raised in the Times piece, and together they raise a bigger question: what if the real obstacle isn’t AI use itself, but the ban-or-allow binary the debate keeps returning to?
The Reframe
The more useful frame, and the one Crompton et al.’s eight areas point toward, is that AI policy is fundamentally an assessment design problem, not just an enforcement problem. Pair academic integrity guidance with integration strategy and human oversight, and the question shifts from “did the student use AI” to “was the assessment built so that AI use is visible, attributable, and still requires the student’s own judgment.” That shift is what lets an online, asynchronous course keep its format, instead of defaulting to a proctored, paper-only fallback that was never designed for how those programs actually run.
This frame is also where the literacy and equity pieces earn their place in the framework, rather than sitting as afterthoughts. Literacy guidance gives students and faculty a shared vocabulary for disclosure instead of a guessing game. Equity guidance keeps a well-intentioned policy from quietly rewarding the students who already have the most access. Neither fixes cheating on its own, but paired with clear integrity rules and real oversight, they replace one fragile rule with a system that can flex from assignment to assignment.
In Practice
This “redesigning the question” frame is how we think about the problem at Collage, an AI-powered teaching and learning platform. Collage is designed around the daily workflows of educators and students: it connects to a course’s instructional materials to help build out its pedagogical structure, supports instructor-defined assessment across formats with supervision capabilities, provides contextual tutoring for students, and gives faculty visibility into where a class is actually struggling. Throughout, faculty stay in control of the learning goals, the rubric, and the final call on every assessment; the platform’s role is to make that oversight practical instead of a bottleneck.
In practice, that means applying instructor-defined rubrics consistently, keeping a record of how a student’s response was evaluated, and surfacing where a class is struggling rather than treating the AI’s read as the last word. That is Crompton et al.’s “human oversight and accountability” area translated into a daily workflow, not a policy document. It is also, we would argue, a more durable answer to the Times piece’s paradox with the return to the blue book: fewer opportunities for undetectable substitution, without asking instructors to give up the modality their course was actually built for.
The institutions making real progress here are not the ones that picked a side of the ban-or-allow debate. They are the ones treating it as eight interlocking design decisions, integrity, ethics, privacy, equity, literacy, integration, oversight, and infrastructure, and building institutional support to hold all eight at once.
See how Collage AI works in practice.
Explore the platform or join a short walkthrough to see how it supports teaching and learning outcomes.
Contact
