A class wants shared rules about AI. I propose an essential pact that specifies objectives, permitted aid, verification of information and how to ask for clarification.
In the example, the document is discussed and updated after a test. The pact does not replace regulations or professional responsibilities.
It helps to make everyday expectations understandable. Its usefulness depends on the possibility of applying it even in difficult cases, avoiding generic formulas that each interprets differently at the time of evaluation.