A professor has accused you of using AI

Answer the message, but do not argue in it

Reply within a day. Silence reads as guilt and often triggers an automatic escalation. But your first reply should be short: acknowledge, state plainly that you wrote it, and ask to meet.

Do not litigate over email. Written exchanges harden positions, and anything you write becomes part of the record. A conversation gives you room to demonstrate that you know your own paper.

Find out what your institution's policy actually says

Before the meeting, read the academic integrity policy. You are looking for three specific things, and students almost never check them.

• Whether a detector score alone can constitute evidence — at many institutions it explicitly cannot
• Whether you have a right to know what evidence is being used against you
• Whether you may bring a support person, adviser or student union representative to the meeting

What to bring

• Version history export, if the document lived in Google Docs or OneDrive
• Outlines, notes, photographed handwritten planning
• Sources you consulted, including ones you rejected — this is very persuasive
• A short timeline of your writing sessions
• Screenshots showing the same passage scoring differently on two or three different detectors

How to talk about the detector without sounding defensive

Do not say the tool is worthless. Say what is true and specific: the score measures how closely writing resembles machine-generated patterns, it produces false positives at a documented rate, and it is known to over-flag non-native English writers and formulaic academic prose.

Then move immediately back to your own evidence. The tool's limitations open the door; your process walks through it.

If it escalates to a formal hearing

Most cases do not get here, but if yours does, the tone changes and so should your preparation.

• Request in writing the full evidence being relied on, including the detector report itself, not a summary of it
• Bring your adviser or union representative — do not attend alone if you are allowed company
• Prepare a one-page written statement and hand it in; do not rely on speaking well under stress
• Ask what the standard of proof is. Academic panels usually work on balance of probabilities, not certainty, which makes your positive evidence more decisive than any attack on the tool

The mistakes that turn a survivable case into a serious one

• Admitting to 'a little AI help' when you used none, hoping for leniency. It converts a contested case into a confessed one.
• Editing or resubmitting the document after the accusation.
• Going public about the instructor on social media while the case is open.
• Missing an appeal window because you were waiting to feel ready.