Loading page
About this tool
Highlight over-used AI phrases like delve, tapestry and testament in your draft, with a plainer replacement suggested for each.
The AI Writing Tell Scanner matches your draft against a bank of 80-plus words and phrases that appear far more often in unedited language-model prose than in ordinary writing — delve, tapestry, testament, realm, seamless, meticulous, plethora, "it is important to note", "in today's fast-paced world" — and highlights each one in place with a plainer replacement. Hits are weighted (strong markers count double) and reported as a density per 1,000 words, so a short email and a long report can be compared on the same scale. The whole scan runs in your browser.
Open AI Writing Tell Scanner 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.
Each hit comes with a concrete plainer alternative you can drop straight in.
Scan only inflated verbs, grand nouns, filler adjectives, stock phrases or hedges.
"Vitality" is never flagged as "vital", so the highlights stay trustworthy.
Delve is a normal English verb, but it appears far more often in language-model output than in ordinary published prose, which made it one of the first widely noticed markers. In most sentences "look at", "dig into" or "study" says the same thing more plainly, so replacing it costs nothing.
No, and that is not the goal. Detection classifiers do not work by word lists, and none of them are reliable enough to be used as evidence. Cutting stock phrasing makes writing clearer and more specific — that is the benefit, regardless of who or what produced the first draft.
No. Each entry is real English used correctly in some contexts — a robust test suite really is robust. Treat a highlight as a prompt to check whether the word is carrying meaning; if a shorter, more concrete word says the same thing, use it.
Every match scores 2 points for a strong marker or 1 for a milder one; the total is divided by the word count and scaled so 25 weighted hits per 1,000 words equals 100. That is roughly one stock phrase every 40 words, or about one every second sentence.