AI writing, measured
Detectors, humanizers and the habits that give AI text away, tested on our own data rather than repeated from someone else's list.
AI humanizer: what it does, and how to judge oneAn AI humanizer takes text a detector reads as machine-written and rewrites it so it reads as a person's. Most fail in one of two ways: they don't move the score, or they move it by changing what you said.Read How to humanize AI textStart with the habits you can see, then fix the ones you can't. We counted which habits really separate AI drafts from human writing, and some of the famous ones don't.Read Can you bypass AI detection?Sometimes, against a specific detector, with a rewrite that changes how the text is built. Here's what moved the score in our tests, what didn't, and why no one can promise it for every detector.Read Do AI detectors flag human writing?Yes, and how often depends on what kind of writing it is and who wrote it. We ran a detector on text written before AI writing was common, so none of it could be AI.Read How do AI detectors work?An AI detector reads your text, gives it a score, and calls it AI if the score crosses a line someone chose. There are three main ways to get that score, and each fails in its own way.Read Signs of AI writing, countedEveryone has a list of AI tells. We counted them in 600 human texts and 900 AI drafts to see which ones hold up, and found they depend on the kind of writing.Read Make an AI LinkedIn post sound like you wrote itLinkedIn is full of AI-drafted posts, and its real posts already read as AI-ish to detectors. That makes it the hardest and most useful place to get right.Read Make ChatGPT text sound humanChatGPT's drafts have a voice people now recognise on sight. The odd part: other models sound so similar that detectors label their text ChatGPT too.Read Make AI-written emails sound like youAn AI-drafted email doesn't fail a detector so much as it fails the reader: it sounds like a template. Most of the fix is cutting, not adding.Read