Loading page
About this tool
Scan an AI draft for tell-tale phrasing and uniform sentences, then work a weighted checklist of edits that restore your own voice.
This checklist measures the surface habits that make an AI draft read mechanically — over-used connectors such as 'moreover' and 'in conclusion', hedge words, em-dash density, -ly adverb rate, repeated sentence openers and low sentence-length variation — then orders fifteen concrete edits so the ones your draft actually needs come first. Sentence-length variation is reported as a standard deviation in words, because uniform sentence length is the loudest single tell in generated prose. It is an editing workflow for writers and editors, not an AI detector, and it makes no claim about who wrote the text.
Open AI Humanizer Checklist on AltFTool — it loads instantly in your browser.
Provide your input — an image, text, or data.
Let the tool analyze or generate the result.
Review, refine, and reuse the output wherever you need it.
Every finding is a number: phrases per 1,000 words, dashes per 1,000, sentence-length standard deviation.
Checklist items whose signal fired in the scan are pinned to the top; the rest sort by impact weight.
The draft is analysed locally in your browser and never uploaded to a server.
Vary sentence length, cut generic connectors, and add one specific detail only you could know. Uniform sentence length is the biggest giveaway — human non-fiction usually shows a sentence-length standard deviation well above five words, while generated prose often sits below it.
The recurring offenders are connector and inflation phrases: moreover, furthermore, additionally, in conclusion, delve into, a testament to, unlock the potential, ever-evolving landscape and in today's fast-paced world. This tool counts more than forty of them and reports the rate per 1,000 words.
No. It measures writing style, not authorship, and it deliberately produces no probability that a text was machine-written. Commercial AI detectors are unreliable and misclassify non-native English writing in particular, so no score of that kind should be used to accuse anyone.
Google's guidance rewards helpful, original content regardless of how it was produced, and penalises mass-produced pages made primarily to rank. The edits here — real numbers, first-hand examples, verified claims — are the ones that add the originality automated drafts lack.