There is no universal rule for acceptable AI assistance. The controlling rules come from the assignment, course syllabus, instructor, department, institution, and applicable data policy. Read them before using AI, and ask the instructor when the boundary is unclear.
Start with the learning objective
AI tutoring supports the student’s learning when it helps the student perform the required thinking. Prohibited substitution occurs when the tool performs work the student is expected to produce independently.
The same action can be acceptable in one course and prohibited in another. A coding instructor may permit an explanation of an error but prohibit generated solutions. A writing course may permit grammar feedback with disclosure. An exam may prohibit all outside tools.
Do not infer permission from technical availability, another course’s policy, or an AI product’s description.
Build a policy hierarchy
Collect the current:
- assignment instructions;
- course AI and collaboration policy;
- syllabus and honor-code references;
- department or institutional academic-integrity policy; and
- privacy, data-handling, or research rules relevant to the material.
More specific authorized instructions normally clarify the task, but do not resolve apparent conflict yourself. Ask the instructor in writing and retain the response.
For a tutoring workflow that keeps the student responsible for the reasoning, see use AI as a tutor, not an answer vending machine.
OpenAI itself advises educators to establish approaches suited to their context and notes that presenting AI-generated content as one’s own may violate local honor codes. See its current educator guidance. The course policy, not this vendor guidance, governs your assignment.
Separate allowed help from substitution
Classify concrete actions:
| Proposed use | Question to ask |
|---|---|
| Request a hint | Does the policy allow tutoring without supplying the answer? |
| Explain an error | May course code or error output be shared with this service? |
| Generate a solution | Is AI-generated work explicitly allowed and how must it be attributed? |
| Review reasoning | Must the initial reasoning be independently produced? |
| Edit prose | Are wording or structural edits allowed? |
| Generate tests | Is test design part of the assessed work? |
Do not label an action “brainstorming” to avoid an inconvenient rule. Describe what information you provided and what the tool produced.
Protect private material
Permission to use AI does not automatically authorize uploading:
- unpublished assignments or solution keys;
- identifiable student information;
- instructor feedback;
- restricted research data;
- proprietary code;
- patient, financial, or employment records; or
- credentials and private repository content.
Managed accounts may have organization-specific administration, retention, and monitoring. OpenAI’s current managed-account notice is one example of why account context matters. Follow the institution’s approved tools and data rules.
Disclose with enough detail
When disclosure is required, record:
- tool and relevant mode;
- date;
- purpose;
- material inputs, with private content omitted;
- nature of output used;
- what you changed;
- verification performed; and
- where the disclosure belongs.
Citation and disclosure are related but not identical. Follow the required citation style and instructor directions. Never invent a transcript or claim independent work when AI supplied material content.
Avoid publishing solutions
A public repository, shared chat link, or portfolio can expose current or future assignment solutions. Permission to submit work does not necessarily include permission to publish the prompt, starter code, test cases, solution, or feedback. Ask before publication, even after the course ends.
Complete the course AI-policy worksheet
# Course AI-Policy Worksheet
Course and assignment:
Authoritative policy locations:
Allowed actions:
Prohibited actions:
Actions requiring disclosure:
Required citation or disclosure format:
Material that must not be uploaded:
Published-solution restrictions:
Unclear scenario:
Instructor question and dated response:
Complete it for one current or fictional course. Use exact policy language only in your private record; do not publish restricted course documents.
Common mistakes
- Assuming tutoring is always allowed.
- Using AI first and reading policy later.
- Treating disclosure as permission.
- Uploading private material to get a better answer.
- Publishing coursework because a repository can be made public.
- Asking AI to interpret an ambiguous integrity rule instead of the instructor.
Do this now
First run a stop gate: confirm the authoritative course rule, permitted kind of help, disclosure requirement, and private-data boundary. If any required answer is unresolved, pause and ask the instructor before using AI.
Complete the worksheet before the next AI-assisted assignment. If any proposed use remains unclear, send the instructor one concrete question and wait for the answer.
Log what you learned
Record only:
- Result: What did the action produce?
- Evidence: What observation, test, or source supports that result?
- Next action or unresolved question: What should happen next?
The next post provides learning-centered prompts to use only inside those boundaries.