Skip to Content
đźš§ This site is under development; its content is not final and may change at any time. đźš§
Teachers' TasksSetting course and assignment AI rules

Setting Course and Assignment AI Rules

Students need to know what AI use is allowed before they begin an assignment. A useful policy connects the rule to the learning outcome, describes allowed and prohibited actions, explains disclosure, and provides an accessible alternative when a tool is required.

A course statement does not replace institutional academic-integrity, accessibility, privacy, assessment, or research policies. Use institutionally approved tools and follow Responsible AI use when a task involves student data or consequential decisions.

Choose the level for each assignment

A single rule for an entire course is often too vague. Choose the level that matches what each assignment is intended to demonstrate.

LevelAppropriate whenExample wording
AI-freeStudents must demonstrate unaided knowledge or performance“Do not use generative AI for any part of this task. The assessment measures your unaided ability to [outcome].”
AI-limitedParticular forms of assistance do not replace the target skill“You may use AI to [allowed actions]. Do not use it to [prohibited actions].”
AI-permittedStudents may choose AI, but remain responsible for the work“You may use AI at any stage, provided you verify the result and disclose how it affected the submitted work.”
AI-integratedUsing and evaluating AI is itself part of the learning outcome“Use the specified AI workflow, retain the required evidence, and evaluate the tool’s contribution and limitations.”

AI-free assessment remains appropriate when unaided recall, writing, calculation, clinical reasoning, performance, or another independent capability is the intended outcome. Requiring AI can be appropriate when critical AI use is an explicit outcome—not merely because a tool is available.

Write a complete assignment rule

Name the learning purpose

State what the student must learn or demonstrate. Explain why the selected AI level protects or supports that outcome.

Define actions, not labels

Terms such as “responsible use” or “AI-assisted” are open to interpretation. Specify whether students may brainstorm, translate, summarize, find sources, generate prose or code, edit language, produce media, obtain feedback, or debug.

Set evidence and disclosure requirements

Ask only for evidence that supports reflection or assessment. A short use record is usually more useful than an indiscriminate transcript or a collection of screenshots. Do not imply that version history or a detector score proves authorship.

Provide access and privacy conditions

If AI use is required, provide access to an approved tool and an equivalent alternative for students who cannot use it because of accessibility, cost, account, location, language, age, or privacy constraints. Do not require students to surrender personal data or rights to complete an activity.

Explain how the work will be assessed

Align the rubric with the learning outcome. Assess the student’s decisions, evidence, reasoning, verification, and reflection where relevant—not the polish of model-generated text.

State the response to uncertainty

Give a contact point and invite questions before submission. Explain that suspected misuse will be handled through the institution’s normal, evidence- based process and that an AI detector score is not proof.

Copyable course statement

Generative AI is addressed separately for each assignment because the permitted role depends on the learning outcome. The assignment brief will label the task AI-free, AI-limited, AI-permitted, or AI-integrated and will list allowed uses. Unless an assignment says otherwise, do not enter personal, confidential, licensed, or unpublished course material into an AI service. You remain responsible for the accuracy, sources, argument, and submitted work. Ask the course team before using a tool when the instructions are unclear. When disclosure is required, briefly record the tool, date, purpose, material retained, and checks performed. Disclosure does not make prohibited assistance, fabricated evidence, or misrepresentation acceptable. Students who cannot use a required tool will receive an equivalent way to meet the same learning outcome without penalty. Contact [role/contact] as early as possible to arrange this.

Copyable assignment block

AI level: [AI-free / AI-limited / AI-permitted / AI-integrated] Learning purpose: [What this task is designed to develop or demonstrate] Allowed: - [specific action] - [specific action] Not allowed: - [specific action] - [specific action] Evidence to submit: [Short disclosure, annotated changes, source checks, reflection, or none] Assessment: [How the rubric rewards the student's knowledge, reasoning, and decisions] Access and alternative: [Approved tool/access route and equivalent non-AI route] Questions or concerns: [contact and applicable institutional policy]

A proportionate disclosure

For many tasks, this is enough:

I used [tool and version, if known] on [date] to [purpose]. I retained [what was used] and checked it by [verification performed]. I remain responsible for the submitted work.

Request a longer process record only when interaction choices are part of the learning outcome. Avoid collecting full conversations by default: they can contain personal information, copyrighted material, false starts, or unrelated content.

Checklist before publishing the rule

  • The AI level follows from the learning outcome.
  • Allowed and prohibited actions are concrete.
  • The rule covers embedded AI features such as translation, rewriting, coding, or image generation where relevant.
  • Disclosure requirements are proportionate and assessable.
  • The institution approves the tool for the intended data.
  • Required use includes access, accessibility, and a genuine alternative.
  • The rubric rewards student learning rather than model fluency.
  • Students know where to ask questions and how concerns will be handled.

References & Footnotes

Last updated on