Guardrails for college instructors

When you finish, you can explain what never goes into an AI tool and how to handle integrity fairly.

Protect student records and work

Grades, student IDs, emails, disability or accommodation information, advising notes, unpublished student work and anything that identifies a student must stay out of consumer AI tools unless your institution has approved that tool for that data. Student-privacy laws exist in many places, such as FERPA in the US.

Student writing may also be the student’s intellectual property. Feeding it into a tool that stores or trains on inputs can raise ethical and legal questions. Ask your institution first.

Know your institution’s policy

Ask your department or teaching center what tools are approved, what the academic-integrity policy says about AI, and what you can decide for your own course. Put your course rules in the syllabus in clear words with examples of allowed and not-allowed uses.

If students must use a particular tool, make sure there is a free or approved option so no one is left out.

Accuracy, bias and sources

AI can invent references that look real, misquote authors and flatten debates in your field. Verify citations. Look at whose perspectives appear in generated examples and whose are missing. Discipline-specific accuracy is your job.

AI detectors are unreliable, and false accusations can harm students, especially those writing in a second language. Do not use a detector score as the only evidence. Use process evidence such as drafts, notes and conversations.

Disclosure and saying no

Model what you ask of students. If you used AI to draft course materials, say so in a short note. Students notice when expectations are one-sided.

Say no to requests to upload class rosters, student work or recommendation files to unapproved tools, and to using AI to make final decisions about grades or misconduct.

A worked example: the suspicious paper

A paper seems too polished. Weak move: run it through a detector and accuse the student. Better: ask the student to talk about the paper, show drafts and explain choices, as your policy allows. Treat it as a conversation about learning, not a trap.

Many students will explain honestly. Some will not. Either way, you have evidence that is human, fair and documented, instead of a number from a tool no one can fully explain.

Practice now

Find your institution’s AI and academic-integrity policy. Write three points in your own words.
Draft three sentences for your syllabus saying how students may and may not use AI, with one example each.

Use any AI assistant you have. Do not paste private, client or patient information into it.

Check yourself

1. Which data must stay out of an unapproved AI tool?

2. A detector flags a paper. What is the fairest step?

3. Why verify AI-generated references?

Ask the AI tutor

Ask anything about this lesson, in any language. The tutor is an AI: it can be wrong, and it never needs your personal details.

Next lesson: Grow with AI: your 30-day plan

AI-drafted. Checked by an automated validator and a separate AI reviewer; a person has not read every lesson. Translations and tutor answers are AI-made. Found a mistake? Tell us. AI can be wrong. General information only, not professional advice. Not affiliated with any AI company named. Current University is not accredited.