Falsely accused of using AI? What to do
Being told your own work was written by AI is stressful, and a detector score is weaker evidence than it sounds. These steps help you respond calmly and with evidence.
General information, not legal advice. Processes differ between schools, employers and publishers. Follow the procedure that applies to you and seek advice from your student union, ombudsperson, HR department or a lawyer if the stakes are high.
First, do not delete anything
Keep every draft, note, outline and file exactly as it is. Do not edit the disputed document and do not backdate anything. Your record of how the work was made is your best evidence.
Steps to take
- Ask exactly what is alleged and what the evidence is. Is it a detector score alone? Ask which tool, what the report says, and which policy or rule applies. Ask for the process in writing.
- Gather your evidence. Version history from Google Docs or from Word files saved to OneDrive or SharePoint, earlier drafts, notes, outlines, source lists, timestamps, and messages with tutors or editors. A document built up over several sessions looks very different from one pasted in whole.
- Be ready to talk through your work. Explain why you made your choices, define the terms you used, expand on an argument, or write a paragraph on the same topic if you are offered the chance. Being able to discuss the work in detail is persuasive.
- Point, politely, to the limits of the tool. Turnitin’s own guidance says AI writing detection may not always be accurate and should not be the sole basis for adverse action against a student. Independent research has found that detectors misclassify non-native English writing more often. OpenAI withdrew its own classifier over a low rate of accuracy. See why detectors produce false positives.
- Respond in writing, calmly and factually. Keep copies of everything you send and receive. Find out what the appeal process is and what the deadlines are, and use them.
- Get support. A student union, academic adviser or ombudsperson, or at work an HR contact or union representative, can help you navigate the process.
What not to do
- Do not run your work through “humanizer” or paraphrasing tools. They change your text, muddy your evidence, may breach the rules, and can make you look guilty of the thing you are denying.
- Do not rewrite the document and resubmit it as if nothing happened. Keep the original as it was.
- Do not rely on a second detector as proof. Checking your writing with another tool, including Litimus, can add context, but a low score from one detector is no more conclusive than a high score from another.
If you are the one making the accusation
A score is a reason to look more closely, not a finding. Read the work yourself, compare it with earlier work, ask about the writer’s process, and give them a chance to respond before you conclude anything. See a fair process for teachers and for writers and editors.
Sources
- Turnitin Guides: Using the AI Writing Report
- Turnitin: Understanding false positives within our AI writing detection capabilities
- Liang et al., “GPT detectors are biased against non-native English writers” (Patterns, 2023)
- OpenAI: New AI classifier for indicating AI-written text (with the July 2023 discontinuation notice)