Text copied from a chat window carries marks readers notice: a chatbot's opening line, Markdown symbols, the same stock phrases. Publisha's editor scores your story for them as you write and clears most of them in one click.
Free on every plan · Scores from 50 words
The editor scores the pasted story 100 out of 100 and lists why: a chat window's opening line and source tag, stock phrases, em dashes and framing lines.
Each rule was tuned on hundreds of real stories published on Publisha, and kept only if it separates pasted chatbot text from writing people typed themselves.
Formatting and wording the editor can correct without changing what you say.
Chatbot leftovers
The opening and closing lines a chat tool adds, source tags copied along with the text, and links ending in a chatbot's tracking code.
Markdown that never became formatting
Asterisks and hash marks that should have been bold text and headings. The fix turns them into the real thing.
A story pasted into a code block
A whole article stuck in a grey box of code text, moved back into normal paragraphs.
Invisible characters
Zero-width spaces and similar characters that travel with copied text and break search and spacing.
Heavy em-dash use
Long dashes between words, far more often than people type them, turned into commas, colons or full stops.
Stock AI phrases
Words such as “delve into” and openers such as “In today's fast-paced world”, swapped for plain ones.
Emoji as bullets or heading decoration
Rows of ticks, rockets and pointing hands at the start of lines, removed.
Patterns only a writer can fix. The editor takes you to each one.
Stock framing lines
Set-up sentences shared word for word by dozens of unrelated writers, such as “This is where X comes in”.
Template headings
Headings built on a chatbot formula, such as “Understanding X” or “The importance of Y”, instead of saying what the section covers.
Placeholder text
Square-bracket gaps such as [insert statistic] that were never filled in.
A heading every few lines
So many headings that the story reads as an outline rather than an article.
The score updates as you write. Next to it is the list of what causes it, each with the passages from your story and what to change.
0
Reads as your own
Nothing in the story reads as pasted from a chat tool.
1 to 29
A few AI patterns
A pattern or two. Often a few em dashes or one stock phrase.
30 to 54
Reads partly AI-written
Enough patterns that readers may notice. The editor asks once before you publish.
55 to 100
Reads AI-generated
Reads as generated. Usually chat leftovers, or several patterns together.
When the score is 30 or more, the editor asks once before you publish: fix and publish, publish as it is, or review first. It never stops you publishing.
Why it matters
People have read enough chatbot text by now to spot it in a sentence. When a story opens with a stock line or is full of the same phrases they have seen a hundred times, many stop reading, and they trust the writer less.
Google has said it judges content on whether it helps people, not on how it was produced, and that using AI tools is not against its guidelines. Its spam policies do target pages produced at scale mainly to rank in search. A story full of generic chatbot patterns looks like that kind of page, even when it isn't. Clearing them out gives your own thinking a fair hearing.
Not an AI detector
AI detectors try to guess whether a person or a machine wrote a text, and they get it wrong often. A 2023 study from Stanford University found that popular detectors flagged a large share of essays by writers whose first language isn't English as AI-written, even though people wrote every one of them.
The AI score does something different. It points to specific patterns you can see and change, shows you where each one is, and explains it. It never labels you, and readers never see the score.
How it works
Open the editor and write, or paste a draft. The score appears once the story passes 50 words.
Under the score, each finding names a pattern and how many times it appears.
Click a finding to select those words in your story, so you can see them in context.
Press the fix button to clear the formatting and stock wording. If you don't like the result, undo it.
Change the framing lines and headings the editor marked, then publish.
Questions
Yes. The AI score and its one-click fix are free on every plan, including the free plan. They run on rules written by Publisha, not on an AI model, so there is no limit on how often you check.
Keep exploring
Paste a draft into Publisha's editor and see its AI score in seconds. The check and the one-click fix are free on every plan.
No credit card · Free on every plan · Undo any fix