Three independent checks before any story reaches your child.
Every storybook on Build Your Book passes through four layers of safety review — keyword filtering, AI classifier, our exclusive subscriber-only Triple-Check, and community moderation. Three of these are fully automated. The fourth is a real human on our team.
How the four layers work
Three independent failure modes have to align before unsafe content can reach a child. Here's exactly how each layer works.
Layer 1
A blocklist that runs before the AI even thinks.
Before any story is generated, your prompt is checked against our hardcoded list of unsafe terms — sexual content, graphic violence, drugs, hate speech, profanity, and child-exploitation signals. If anything matches, generation never starts.
Deterministic regex match · zero false-negatives on listed terms · 0 ms cost · catches things at the gate.
Layer 2
A second-pass AI that reads what was written.
After our story-writer AI finishes, a separate moderation model re-reads the full story to catch anything implicit — scary themes, unresolved conflict, fear, separation, or language that's a bit too old for your kid. Borderline content is hidden until a human admin approves it.
Gemini 2.5 Flash classifier · scores every chapter independently · "needs_review" verdicts route to admin queue, never auto-publish.
Layer 3
The Triple-Check — our subscriber promise.
For our paid plans, every story gets one more pass through a strict 'sensitive grandparent of a 3-year-old' AI persona. Its only job is to flag ANYTHING that could upset, scare, or confuse a young child — including subtle things our standard checks might miss. If it flags the story, we automatically regenerate with a softer prompt.
Independent prompt + persona from Layer 2 · single-word verdict · automatic one-shot regeneration on UNSAFE · books that pass earn the green Verified Safe badge.
Layer 4
You and our community are the final check.
Anyone can flag a book with one tap. After 3 reports, the book auto-hides from the marketplace pending review. Our admin team manually reviews every flagged book and every book in the moderation queue. We learn from every report and update our blocklists weekly.
Automatic hide threshold = 3 reports · admin moderation queue with one-click decisions · weekly blocklist updates from learnings.
The green Verified Safe badge
When you see the green ✓ Verified Safe badge on a book, it means that book passed all four safety layers AND our exclusive Triple-Check audit. It's the strongest safety guarantee we offer — and it's available on every book generated by an active paid subscriber.
Beyond content — your privacy too
Zero data resale.
We never sell or share your child's reading data with anyone. Period.
Encrypted in transit.
All data flows over TLS 1.2+. Passwords are bcrypt-hashed; sessions are httpOnly cookies.
You decide what they see.
Every parent account can set an age range. Books outside that range never appear in search results for that account.
Ready to start a story tonight?
Every story you create or buy on Build Your Book is triple-checked, child-safe, and ready to be tomorrow's favourite bedtime memory.