Widely available AI writing tools have changed how teachers think about take-home essays, and a well-known topic like the September 11 attacks is especially exposed. A generic prompt about the causes of the attacks can be answered by a chatbot in seconds, producing fluent but shallow text. Assignments built on Sid Jacobson's The 9/11 Report can be designed to require engagement with specific details that general tools handle poorly. Clear policies and thoughtful prompts protect both learning and fairness.

Start with a clear statement of what is and is not allowed. Students need to know whether they may use AI to brainstorm, check grammar, or summarize, or whether any use at all is prohibited. Ambiguity breeds accidental violations and unfair accusations. Put the policy in writing, discuss it in class, and ask students to acknowledge it with each submission.

It also helps to explain the reasoning behind your rules. When students understand that the goal is to practice analyzing evidence themselves, they are more likely to respect the boundary. A policy framed only as punishment tends to invite workarounds. Honest conversation builds a culture of integrity.

Designing Prompts That Require Real Engagement

Prompts that ask for personal analysis of specific pages are harder to outsource. For instance, asking students to analyze how a particular sequence of panels conveys the pace of events requires them to have read and looked closely at the book. Requiring references to scenes, captions, or visual choices makes generic output easy to spot. These prompts also tend to produce more interesting essays.

• Ask for analysis of specific chapters, scenes, or panels
• Require students to cite page locations or describe images precisely
• Include an in-class writing component to establish a baseline voice
• Use process checkpoints such as outlines, notes, and drafts
• Add a brief reflection explaining how the student developed their argument

Evaluating Suspicious Work Responsibly

Detection software is unreliable and can produce false accusations, so avoid treating its output as proof. A more defensible approach looks at the work itself, including inaccuracies about the book, a sudden change in style, or an absence of the specific details the prompt required. Follow up with a conversation in which the student explains their argument and sources. This gives them a fair chance to demonstrate authorship.

Document your process and follow your school's policy. If you have concerns, gather the evidence calmly and involve administrators as required. Remember that a student's English proficiency or writing growth can change style without any misconduct. Fairness requires humility about what you can conclude from text alone.

Teaching Responsible Use

Rather than only policing, consider teaching students how to use AI appropriately. A lesson might show how to ask an AI tool for questions to guide research while still writing the analysis independently. Students can also practice checking AI output against the book to see where it errs. This builds critical thinking and digital literacy.

Evidence of inaccuracy is a powerful teacher. When students discover that a chatbot misattributes events or invents details about the Commission's work, they learn to verify. That lesson serves them well beyond this unit. It connects directly to the historian's habit of checking sources.

Using Technology on the Teacher Side

Teachers can use AI-assisted grading tools to give consistent, rubric-based feedback, freeing time for more conferences and in-class writing. Be transparent with students about how such tools are used and who reviews the results. Trust grows when expectations apply to both sides. The aim is to strengthen learning, not to automate judgment.

Revisit your policies regularly, since tools and norms change quickly. Gather feedback from students and colleagues about what is working. A flexible, thoughtful approach will serve better than a rigid rule set. With clear communication and well-designed assignments, integrity and innovation can coexist.