Academic integrity and AI tools: a plain guide to staying within the rules

We make a tool students use, so we'll be direct about where the line is. This is a plain guide, not legal advice, and your institution's policy always wins.

Read your policy first

AI policies range widely: some universities allow AI for brainstorming, outlining, and language support with a declaration; some allow it for everything with citation; some ban it for assessed work entirely; many are still being written. You cannot reason about whether a use is acceptable without knowing which regime you're under. Find your institution's policy, and if it's unclear, ask your instructor in writing.

Where disclosure is required, disclose. A humanized draft is still AI-assisted; making it read naturally does not change what you used to produce it, and passing it off as unassisted where the rules require disclosure is a violation regardless of how it reads.

Staying the author

The safest and most defensible workflow is the one where you remain genuinely the author: you decide the argument, you understand and can explain every sentence, and you verify every claim against real sources. AI can help you draft and language tools can help you polish, but if you couldn't reconstruct the reasoning in a viva, the work isn't yours yet, and no tool fixes that.

Used this way, a humanizer is an editor: it removes the machine's tell-tale evenness from sentences you already understand and stand behind. Used to disguise work you didn't do, it's a way to break your institution's rules, and we'd rather you didn't.

Why 'will it pass?' is the wrong question

Detectors are estimates that misfire, so 'will this pass the detector' is both unanswerable and beside the point. The question that keeps you safe is 'can I stand behind this, and have I followed my institution's rules, including disclosure?' If the answer is yes, a detector's noisy guess is a problem to resolve with evidence of your process, not something to pre-empt by gaming a score.

For non-native English speakers

If English isn't your first language, AI assistance and language tools are legitimate supports that your policy most likely permits. Ironically, the polishing they do can raise a false-positive risk. Keep your drafts, follow your disclosure rules, and use a humanizer to restore your natural voice, not to hide legitimate help.