Community Guidelines

Describe the manuscript you are sharing

• Human-written. You wrote the text. Correcting spelling and grammar, dictating your own words or using accessibility tools fits this category. Example: “Human-written. I dictated the draft and used a grammar checker.”
• AI-assisted. AI helped plan the story or substantively edit it, but you wrote the text and did not retain AI-drafted passages. Example: “AI-assisted. I used AI to discuss the outline and suggest structural edits; I wrote the text.”
• Contains AI-generated text — some or most/all. If AI drafted passages you kept, choose this category even if you edited those passages later. Say whether some or most/all of the text is AI-generated; any further explanation is optional. Example: “Contains AI-generated text — some. AI drafted two passages that I revised and retained.”

A clearly identified quotation showing an AI example does not make your surrounding writing AI-generated. For example, an essay you wrote that labels a quoted chatbot response as an AI example can still be human-written.

Disclose AI cover images and AI translations separately. For example: “Manuscript: Human-written. Cover: AI-generated. Translation: AI-translated from my human-written original.” A human-written original does not remove the need to disclose an AI translation.

Make the disclosure visible and keep it current

These labels are your declarations, not verification by BetaReader. They describe the version readers receive. Update the declaration when that version changes. Until dedicated controls are available, put disclosures clearly at the start of the description, visible before readers join. Until permission controls are available, ask the author directly through the community process and obtain separate explicit permission before sending manuscript text to an external AI service. That permission cannot allow AI-produced reader feedback.

An existing work without a declaration is not presumed human-written. Existing Discovery listings will receive an in-app notice and have 30 days from the first delivery of that notice to declare. The clock does not start on 18 October 2026 just because these guidelines were updated.

Be honest about the feedback

• Your reading, your words. Reader comments, critiques, reviews and narrative survey responses must be based on your own reading and written in your own words. Example: you read a chapter and write your own reaction to its pacing.
• Limited tools are allowed. Spelling correction, dictation, accessibility tools and faithful translation of your own words are allowed only if they do not generate observations or rewrite the feedback. Example: you dictate your own reaction or faithfully translate your own comment without adding analysis or rewriting it.
• No AI-produced reader feedback. Do not submit AI-written feedback, AI-assisted rewriting or AI-generated analysis, even if the author gives permission. Example: asking AI to rewrite your chapter notes as a critique is not allowed, even when the notes are yours.

The dedicated app confirmation checkbox will ask: “I confirm this is my own feedback, based on my reading, and was not written or generated by AI.” Until it rolls out, the same rule applies.

BetaReader's own AI editor and digest reports must be explicitly labeled as generated and kept separate from reader submissions. An AI digest summarizing human comments does not turn the original comments into AI-generated feedback. Neither this policy nor the confirmation certifies or relabels historical feedback records.

Ask separately before using external AI services

Being allowed to read a manuscript is not permission to send manuscript text to an external AI service. You need separate, explicit author permission first. That permission cannot override the prohibition on AI-produced reader feedback.

Example: permission to read a book does not allow you to paste a chapter into an external AI service. Ask separately before sharing manuscript text; even with that permission, you cannot submit the service's output as reader feedback.

Your own notes that contain no manuscript text, identifying details or private details are distinct from uploading a manuscript. For example, a note in your own words about pacing is different from a pasted chapter or private character details. That distinction does not allow AI to rewrite the note for submission as reader feedback. These disclosures and permissions grant no model-training rights. Authors and readers remain responsible for what they submit, truthful declarations, rights and confidentiality.

Keep Discovery recruitment voluntary

Invite people to give voluntary manuscript feedback. Do not require a purchase, payment or an external public review, and do not send spam. “Would you like to read my draft and share your observations?” is appropriate; “Buy my book and post a public review to join” is not.

Manuscript comments, critiques, surveys and ratings are not the product-improvement “Feedback” covered by section 8 of the Terms. Suggesting a new BetaReader feature is product-improvement feedback; commenting on an author's chapter is manuscript feedback.

Report misleading disclosures fairly

Send a book or feedback link and the reason for your concern to hello@betareader.io. We request corrections first where appropriate and escalate repeated deception. An automated detector result is not an automatic verdict.

For example, a missing or outdated declaration may need a correction. Repeated misleading declarations after correction requests can be escalated. Do not treat a detector score alone as proof that someone has misrepresented their work.