Frameworks to borrow from
Current frameworks and recent research first, then foundational frameworks that still apply.
- Frameworks are complementary: each answers a different question.
- Combine an assignment-level scale with institutional and inclusive-design frameworks.
- Author-developed frameworks are labeled as such.
Frameworks compared
These frameworks are complementary, not competing. Each was built for a different purpose and unit of analysis, so most institutions combine several.
How much AI may students use on this task?
- Use it to
- Set and communicate AI use per assignment.
- Audience
- Instructors, students
- Strength
- Simple shared vocabulary for permissions
- Limitation
- Says how much AI, not why or whether learning improves
What should university GenAI policy cover?
- Use it to
- Check a campus policy against eight expert-agreed areas.
- Audience
- Policy makers, leaders
- Strength
- Global expert agreement on policy scope
- Limitation
- Expert consensus, not tested outcomes
How do we keep AI use human-centered and inclusive?
- Use it to
- Ground policy in human agency, inclusion and privacy.
- Audience
- Governments, institutions
- Strength
- Human rights and inclusion focus
- Limitation
- Broad; needs local translation into practice
How do we govern, map, measure and manage AI risk?
- Use it to
- Review tools and uses before approving them.
- Audience
- IT, risk and governance teams
- Strength
- Structured, widely used risk process
- Limitation
- Not designed around teaching and learning
Can every learner engage and show learning?
- Use it to
- Offer more than one route, with and without AI.
- Audience
- Instructors, designers
- Strength
- Strong base for inclusive course design
- Limitation
- Predates generative AI; AI uses must be inferred
Which trends are reshaping teaching and learning?
- Use it to
- Plan for change in assessment and instructional design.
- Audience
- Leaders, teaching centers
- Strength
- Current view of sector trends
- Limitation
- Forecasting, not evidence of what works
Which documents does AI governance need, and who sees them?
- Use it to
- Connect classroom guidance to approved tools and reviews.
- Audience
- Governance and IT teams
- Strength
- Links documents, tools and visibility
- Limitation
- Author-developed; not yet independently evaluated
Three kinds of source in this guide
Research evidence. Peer-reviewed studies, working papers and surveys, labeled by type on the Evidence page.
Institutional practice. Published university policies and syllabus statements, cited as guidance in force, not as proof of effectiveness.
Author frameworks. The Campus AI Registry and the learning-first model are the author’s own and are labeled as such.
Three tiers of policy
Recent research finds instructors acting as gatekeepers within adaptive institutional frameworks (Studies in Higher Education, 2026). Foundational work: the MLA-CCCC Task Force (2024) and Chan (2023) first proposed these tiers.
Institution
Baseline values, a default rule for silent syllabi, data privacy, approved tools and integrity procedures.
Program
Discipline norms, shared AI levels for common assignment types, and consistent expectations across sections.
Course
The syllabus statement, an AI level for each assignment, disclosure requirements and a conversation with students.