Colleges Are Scrambling to Write AI Policies-And Instructors Are Split on How to Enforce Them
Forty-five percent of colleges and universities had adopted AI policies by 2025, up from 20% the year before. These policies typically cover academic integrity, teaching, research, or combinations of all three. Most institutions give broad guidelines and leave enforcement to individual instructors.
The variation is stark. Some instructors prohibit AI use entirely. Others encourage it. Some institutions mandate proprietary AI tools for privacy reasons, while many now require instructors to include an AI policy in their syllabi.
Setting Clear Expectations From Day One
Instructors who succeed tend to be transparent about their own AI use and involve students in policy creation. One approach: disclose your own practices on the syllabus-for example, "I use AI to develop course materials but not for grading"-then work with students to draft a classroom policy collaboratively.
This shared ownership matters. Policies drafted with student input typically permit AI for brainstorming, research, and feedback but prohibit it for generating finished prose. Students are more likely to follow rules they helped create.
Some instructors go further and let students create personal AI policies based on their own ethical concerns about privacy, bias, or environmental impact. For those who decline AI use, instructors offer alternative assignments that don't require it.
Detection Tools Aren't Reliable-And Instructors Know It
Nearly 80% of U.S. faculty said cheating has increased since AI tools became widely available. About half use detection software like Turnitin or GPTZero. Yet only 1% said they completely trust these tools.
The reasons are concrete. AI detection software disproportionately flags original work from neurodivergent students and those writing in a second language. Research also shows instructors overestimate their ability to spot AI-generated essays.
Relying on detection tools creates another problem: false accusations erode trust. Instructors report that using checkers shifts their role from teaching to policing.
An Arms Race With No Winner
As detection tools improve, "humanizer" software emerges to evade them. This cycle echoes past education battles-the calculator wars in mathematics, for instance-that technology ultimately won.
Rather than chase detection, some instructors focus on accountability. The core message: whoever submits work owns it. If AI plagiarized, the student plagiarized. If AI made an error, the student made an error.
This approach shifts responsibility away from catching cheaters and toward teaching students to think critically about their choices with AI, both in your classroom and beyond.
For guidance on creating classroom policies, resources like the AI Pedagogy Project and Harvard's generative AI guidance offer sample syllabus language. The most effective policies tend to be the ones instructors and students build together-not the ones imposed from above.
Your membership also unlocks:
