Creating a System to Prevent Cheating with AI

(Note: This is a recording of the presentation used during the November 2025 webinar. Newsletter attendees receive invitations to the live webinar.)

TL;DR: Preventing Cheating With AI While Strengthening Learning

Overview

AI makes it easier than ever for students to outsource schoolwork, but cheating is not new and cannot be prevented through prohibition alone. The most effective response is to redesign policies, assignments, and assessments so the focus remains on learning, and so students have fewer reasons or opportunities to cheat.

This summary distills the key ideas and instructional strategies for creating an environment where students learn deeply and demonstrate their own knowledge, even when AI tools are readily available.

Clarifying Cheating and the Role of AI

Cheating occurs when students present work or ideas that are not their own without attribution. Because AI tools can generate polished products instantly, students can easily misuse them unless expectations are clearly defined. Policies must explicitly describe what counts as cheating, how AI may be used, and what students must disclose about their AI use.

Rather than banning AI, educators should recognize that students will use it and instead design learning experiences that highlight thinking, application, and personal engagement, which AI cannot replicate.

Foundational Conditions for Reducing Cheating

1. Clear AI and Academic Integrity Policies

Policies should define cheating, establish expectations for citation, and outline structured levels of permitted AI use. Students need consistent guidance so they understand what is allowed, what is not, and why.

2. Structured Levels of AI Use

The five levels of AI use, ranging from no AI to full co-creation, help teachers match appropriate AI use to assignment goals. Students can then understand which uses support learning and where the boundary of misuse begins.

3. A Shift in Assessment Philosophy

Effective assessment prioritizes what students know over how polished their products appear. Educators should emphasize mastery of knowledge, regularly use low-stakes formative assessment, and distinguish between learning activities done at home and assessment activities completed in supervised settings.

4. Assignments That Highlight What Only Students Can Do

AI cannot reason about lived experiences, justify thinking, build relationships, or apply the “smell test” for plausibility. Assessments should draw on these uniquely human abilities so students must rely on their own understanding rather than AI output.

Assessment and Assignment Strategies That Are AI-Resistant

1. Performance-Based Strategies

Students demonstrate skills or knowledge directly in front of the teacher. Examples include oral defenses, short-answer tasks, demonstrations, station rotations, and problems requiring hands-on manipulatives.

2. Application and Transfer Strategies

Students apply learning to new or personal contexts. Approaches include case studies, real-world scenarios, reframing knowledge for new audiences, and reading passages with targeted response questions.

3. Production Strategies

Students create tangible artifacts that reveal thinking, such as curated portfolios, lab notes, journals, quick-writes, or student-generated problems and solutions.

4. Metacognitive Strategies

Students explain and justify their reasoning. Examples include reflections, short conferences, debates, seminars, and written justifications. These depend on personal thought processes that AI cannot supply.

5. Iterative or Multi-Stage Strategies

Students work through drafts, checkpoints, and revisions with teacher oversight. Because the teacher observes each developmental step, AI cannot generate the full sequence for them.

Key Takeaway

Preventing cheating with AI does not depend on stronger policing. It depends on designing learning and assessment experiences that emphasize what students know, require human judgment and explanation, and give no advantage to outsourcing work to AI. When assignments prioritize thinking and application rather than polished products, students demonstrate genuine understanding and have far fewer opportunities or incentives to cheat.