Glyphtell Research Prototype

What counts as cheating with AI?

There is no single rule. What is allowed depends on the course, the assignment and sometimes the individual instructor. Here is how to find out where the line is.

Last reviewed 12 November 2026

A great many AI misconduct cases involve students who did not think they were cheating. One instructor encourages chatbots for brainstorming; the next treats a grammar tool as unauthorised help. The same action can be fine on Monday's assignment and a violation on Friday's.

Where the rules actually are

Check these in order. The more specific source normally wins.

  1. The assignment brief. Any statement about tools on the task itself.
  2. The course syllabus or handbook. Look for a section on AI, academic integrity or permitted resources.
  3. Department or programme rules.
  4. The institution's academic integrity policy. If it says nothing specific about AI, older rules on "unauthorised assistance" or "work that is not your own" usually apply.

If none of these answers your question, ask the instructor by email before you start, and keep the reply.

A rough spectrum

These are generalisations. Your course may draw the line anywhere along this list.

Usually accepted

Depends on the course

Usually treated as misconduct unless expressly allowed

Invented references are a separate problem. Chatbots can produce citations that look real and do not exist. Submitting them can be treated as fabrication, which many policies regard as more serious than unauthorised help.

Three questions that settle most cases

Protecting yourself

Already in trouble over something in the grey area? Describe exactly what you did and let the instructor judge it. See Should you admit to using AI?