Our AI promise
You write your book.
BlurbBio uses AI to read your manuscript, point out problems, teach and suggest. Here is exactly what it may do with your writing, what it never does, and how you can check for yourself.
You write your book
No BlurbBio feature writes chapters or scenes of your manuscript. AI text goes into a chapter only when you click to accept a suggestion: a rewrite of a passage you selected, a suggested word, a fix, a summary from a generator. Nothing is added on its own.
Some tools draft text around your book when you ask, such as a blurb, a synopsis or an outline. They never put it into your chapters.
Every AI suggestion you accept is marked
Accepted AI text carries an invisible mark in your manuscript, so you can see it in the chapter and find it later. Your book’s AI disclosure report in Launch Kit lists each marked passage, and you can download an authorship record: everything BlurbBio’s AI did with your book, in one file to keep or show a publisher.
What it can’t see: text you paste in from elsewhere, AI suggestions accepted before marking began (the report tells you if your book has any), single names from the name generator, and rule-based spelling fixes.
The Practice Room coach never writes for you
The coach marks your exercise against a rubric you can read before you start, quotes your own words as evidence, and tells you the one thing to work on first. It never rewrites your piece. When you put a revised passage into your chapter, the words are yours.
We don’t train AI on your writing
BlurbBio does not train AI models, and never sells or hands over your writing for anyone to train on.
To do what you ask, your text is sent to the service that does the work:
- OpenAI runs BlurbBio’s AI features. Its terms say data sent through its API is not used to train its models unless the customer opts in (BlurbBio has not), and is kept for up to 30 days for abuse monitoring. OpenAI’s data terms.
- LanguageTool checks spelling and grammar in Deep Check. Its policy says it does not use users’ text to train its models. LanguageTool’s privacy policy.
Feedback you can check
Every Practice Room rubric is shown to you in full: what Not yet, Getting there and Solid look like. The coach can misread a piece, and you are always the judge. We are asking professional editors to mark a sample blind, and we will publish how often they agree with the coach, including where it doesn’t.
Last checked 17 November 2026. If a feature changes what AI can do with your manuscript, this page changes first. Questions: see our privacy policy or contact us.