La Gloire de mon père is a short, popular book with a wealth of online summaries, and a student can now generate a competent-sounding essay about it in under a minute. Teachers are right to worry, but banning tools rarely solves the problem and detecting them is unreliable. A better approach is to design assignments that make student thinking visible and to state clear expectations about what help is allowed.

Begin with a written policy that students can actually understand. It should say whether AI may be used for brainstorming, for language checking, or not at all, and it should explain the reasoning in plain language. Ambiguity breeds accidental violations, and students are more likely to follow rules they understand.

Then consider what makes an essay on this book difficult to fake. Prompts tied to specific class discussions, to particular page references, or to a student's own chosen passage are harder to answer with generic text. A prompt asking how a scene from the hunting chapter connects to a claim made in class requires information a chatbot does not have.

Build Process Into the Assignment

Collecting work at several stages makes the writing process visible and makes it harder to submit an unfamiliar finished product. A brief outline, an annotated passage, a rough draft, and a short reflection each reveal the student's thinking. Teachers who see the progression can usually tell whether the final essay grew out of the earlier work.

• A handwritten or in-class annotation of a key passage
• A one-paragraph thesis proposal with a chosen quotation
• A rough draft submitted on a set date
• A short reflection explaining what changed during revision
• A brief conversation or oral check on the finished essay

Use Grading Criteria That Reward Specificity

Generic essays tend to share a pattern: broad claims, safe themes, and few precise references. A rubric that rewards accurate, specific textual evidence and original interpretation naturally scores such papers lower. That approach shifts the focus from policing to teaching what good analysis looks like.

Be careful with accusations. Detection software produces false positives, and an incorrect accusation can damage a student's trust and record. If you suspect a problem, start with a conversation in which the student explains their argument, and let the quality of that explanation guide your next step.

Teach Responsible Use Openly

Students will encounter AI throughout their lives, and teaching them to use it responsibly is part of education. Show them how to use a chatbot to check a French sentence or to brainstorm questions about the memoir, while insisting that the argument and the evidence come from their own reading. Naming the boundary clearly makes it easier to respect.

Teachers can also use AI in a transparent way themselves. GraideMind, for example, applies a teacher's rubric to produce draft feedback that the teacher reviews, and explaining that practice to students models the idea that tools can support human judgment without replacing it.

Revisit the Policy Regularly

Tools and student habits change quickly, so a policy written once will soon be out of date. Review it at the start of each term and adjust the language based on what you observed the previous year. Inviting student input can also surface practical concerns you had not considered.

Share the updated policy with families and colleagues so expectations stay consistent across classes. A department-level approach prevents students from receiving conflicting messages and makes enforcement fairer.