Requests
Illegal, exploitative, sexual, hateful, dangerous, impersonating, or privacy-invasive requests are rejected. Child likeness uploads and voice cloning are outside V1.
CONTENT SAFETY
A layered review before the family sees a story.
Hadoty layers recipe constraints, deterministic checks, model review, adult preview, reporting, and operational kill switches.
Story review
Drafts are checked for age-appropriate language, prohibited content, harmful instructions, character continuity, selected language, and the intended value.
Illustration review
Image prompts exclude identifiers and are moderated separately. Text remains available when an image cannot be safely produced.
Escalation
Hard failures enter a held state. Adults can revise, leave private, or report; authorized operations can pause a model or capability without releasing the draft.