Content policy
Version 2026-09-04 · also available as JSON at /content-policy.json
LoreChat is a storytelling workspace for character cards, and everything on it is fiction. It is fade-to-black: romance, attraction, tension, kissing, implied intimacy, mature themes and fictional violence are allowed. Explicit sexual description is not, anywhere on the service, in any language.
Allowed
- Romance, attraction and flirting
- Slow burns, enemies to lovers, confessions, jealousy, heartbreak.
- Tension and desire
- Longing, chemistry, charged silences, a hand that lingers.
- Kissing and embraces
- Kisses, holding, dancing close, falling asleep together.
- Implied intimacy
- A scene that cuts away at the bedroom door and picks up the next morning. What happened is understood, not described.
- Mature themes
- Grief, addiction, crime, betrayal, trauma, moral grey, dark humour, strong language.
- Fictional violence
- Fights, battles, injury, death, horror, war and their aftermath, within a story.
- Non-sexual nudity
- Bathing, medical scenes, life drawing, described without sexual detail.
Not allowed
- Explicit sexual content
- Descriptions of sexual acts, genitals, or bodies in a sexual way, in narration or dialogue, and instructions telling the model to write them. This applies to your messages, character replies, cards you import, create or publish, personas and memory notes, share links and illustration prompts.
- Sexual content involving minors
- Anyone under 18, anyone described or coded as a minor, and age-play framing. No exceptions.
- Non-consensual sexual content
- Sexual content involving force, coercion, assault, drugging or an inability to consent.
- Illegal sexual content
- Bestiality, incest, necrophilia, sexual torture, and sexual content tied to trafficking.
- Sexual content about real people
- Celebrities, public figures, streamers, or anyone real, in sexual scenarios.
How it is enforced
| Where | How |
|---|---|
| Card import, create, edit and publish | Every field, tag and lorebook entry is checked before it is saved, and the reason names the field that failed whenever the match sits in one field; a combination spread across fields is refused without naming one. Cards labelled NSFW by their source, and cards that instruct the model to write explicit content, are refused. Publishing runs a second, model-based check and will not publish while that check cannot run. |
| Your messages | Every message is screened against this policy before it reaches the model, including common obfuscations such as spaced-out letters and character swaps. A message the screen catches is refused and never sent. |
| Character replies | The model is instructed to fade to black. Every reply is checked before it is delivered; a reply that crosses the line is discarded and the credit refunded. |
| Personas, memory notes and story summaries | These feed back into the prompt every turn, so they are checked on every save. |
| Share links and public pages | A share is checked when it is created and again every time it is served. A public card is checked when it is published and again when it is served. |
| Illustrations | Scene prompts keep everyone clothed and are checked before an image is generated. |
| Records | Every block at a write boundary — a message, a reply, a card, a persona or memory note, a share, an illustration prompt — is logged with the surface, the rule that fired, a SHA-256 of the text (never the text itself) and the policy version, so enforcement can be audited. Public pages are re-checked on every read; those read-time blocks are not written to the log, because crawler traffic would dominate it. |
Reporting
Anything that slips through can be reported from a reply's flag menu or from the Report link on a share page.
Who can create an account is covered by the Terms of Service.