Should Schools Ban AI Humanizer Tools? The Honest Case for Both Sides
As academic integrity lead at my school, I’ve been asked to recommend a position on AI humanizer tools. Before I do, I want to understand both sides better.
The case for banning them: their primary use case is bypassing AI detection. Unlike ChatGPT which has legitimate educational uses, humanizers exist specifically to make AI-generated text undetectable. That’s the product. Banning them sends a clear signal about academic honesty.
The case against banning them: enforcement is essentially impossible. Students access these tools through personal devices on personal networks. A ban with no enforcement is worse than no ban – it communicates rules you don’t intend to keep. Also, some uses (improving AI-assisted drafting that the student disclosed) may be legitimate.
The middle position: require AI disclosure, teach students what humanizers do and why they’re ethically problematic, and build assessments that make them irrelevant rather than trying to ban tools that can’t be banned.
what’s your school actually doing?
6 Replies
Join the discussion.
Log In to Reply
wish our admin would read this thread honestly. we went straight to "ban QuillBot" without the conversation you're having here. students didn't stop using it - they just didn't tell us anymore. the ban achieved secrecy, not compliance.
Katherine's middle position is where our board landed after extensive consultation. The logic: Grammarly is already a widely accepted writing aid. The line between Grammarly and an AI humanizer is a matter of degree, not kind. Rather than drawing an unpoliceable line, we require disclosure of any AI assistance and focus assessment on demonstrated understanding rather than text authenticity.
from an ethics standpoint: the question of whether humanizer tools are deceptive depends entirely on what the student is representing. submitting AI-generated content as original human writing is deceptive regardless of whether it's been humanized. the humanizer doesn't create the ethical problem - the misrepresentation does.
the enforcement problem with banning humanizer tools is that they're indistinguishable from good editing and revision. you ban humanizers and you functionally punish any student who revised heavily. the rule is unenforceable and the attempt to enforce it will produce exactly the false positive/equity problems weve already seen with standard AI detection.
ran a full semester ban experiment. students got sneakier. thats it. thats the whole result. transparency model is the only thing that actually changes behavior because at least then you're in the conversation.
same result in my school. and sneakier is actively worse than open use because at least with open use you're still in the conversation. the ban removed the conversation entirely. students stopped asking questions about AI, stopped disclosing anything, and just worked around it silently.