Safety Policy

close up of hands taking notes and several people at a table

Cantina is an 18+ platform. This page sets out what our policies cover, how we enforce them, and how to reach us when something goes wrong. The detail lives in our Community Guidelines, which are published as individual policies so each one can be specific.

Sitting at the intersection of social and AI means intent and context matter. The same prompt can be a joke in one room and a targeted attack in another, so our policies are written to be read against context rather than applied mechanically, and we revise them as new patterns of misuse appear.

Community Guidelines

Our Community Guidelines set out what is allowed and what is not. They are deliberately permissive about creative range — weird, dark and provocative work is welcome here — and deliberately narrow about the categories that carry real risk of harm. The sections below are the priorities those guidelines are built around.

View our full Community Guidelines as well as our full Terms of Service.

Policy Priorities

Child Safety

There is no grey area here and no creative exception. Any content that sexualises, exploits or endangers anyone who is or appears to be under 18 is prohibited, whether it is real, fictional or AI-generated, and whether or not a character is described as an adult. Every instance of child sexual abuse and exploitation is reported to the National Center for Missing & Exploited Children, where Cantina is a registered electronic service provider. Known abusive material is detected by PhotoDNA against hash lists we receive under agreement with NCMEC. This category is enforced more strictly than any other, including against accounts otherwise in good standing.

View our Child Sexual Abuse & Exploitation policy

Age Requirement

Cantina is for adults. You must be 18 or over to hold an account, and we keep checking after sign-up. If a date of birth, an automated signal or a moderator review suggests an account may belong to someone under 18, that account is locked pending verification. Verification is a short video selfie, with a government ID scan as fallback if the selfie is inconclusive. It is processed by our partner Cinder using FaceTec technology; Cantina receives only the result and a reference number, never your selfie, your ID image, your face geometry or any biometric identifier. If verification confirms the account holder is under 18, the account is permanently suspended.

Read how age verification works

Non-consensual Intimate Media

Zero tolerance. Generating or sharing intimate or sexualised content of a real, identifiable person who has not consented to it is prohibited, and that includes “undressing” or “nudify” prompts applied to anyone’s photograph, public figures included. Where the source image came from makes no difference: a public profile, a press photo or an image someone sent you are not consent. Content is removed and matched against the StopNCII and Take It Down hash lists, so imagery registered with those services is blocked here too.

View our Non-consensual Intimate Media policy

Suicide and Self-Harm

Content that encourages or gives instructions for suicide, self-harm or disordered eating is prohibited. If a conversation with a character turns to suicide or self-harm, the character stops and Crisis Text Line’s dedicated Cantina number is shown automatically, and a person reviews what happened afterwards.

View our Suicide and Self-Harm policy

Responsible Use of AI

Prompts are screened before anything is generated and the output is screened again before it reaches anyone. Media generated on Cantina carries C2PA Content Credentials, and anything shared off the platform carries a visible watermark, so where something came from travels with it. Characters cannot depict minors, use a real person’s likeness without consent, or claim professional credentials they do not hold.

Intellectual Property

Users must respect copyright and trademark law. We provide in-product education on intellectual property and follow DMCA notice and takedown procedures, and our DMCA policy covers all content on Cantina including AI-generated images, video, audio and avatars.

If your copyrighted work has been used without permission, submit a takedown request using our DMCA form. If instead your name, image, voice or likeness has been used without your permission, that goes through our Likeness Policy, not DMCA. Complaints involving minors, private individuals or intimate imagery receive expedited review.

Cantina Safety - photo of people's hands working at a table and one woman at a white board

Collaboration and Transparency

Context is most of this work, so our product and Trust & Safety teams keep regular feedback loops with the community. We also sit inside the organisations that set standards for this field — the Technology Coalition, the Digital Trust & Safety Partnership and the Family Online Safety Institute — on working groups, research councils and shared commitments, because the abuse patterns in social AI are new and nobody works them out alone.

Our Community Guidelines set out what is and is not allowed, and why. We hold back some enforcement detail such as thresholds, detection signals, the specifics of how a classifier is tuned, because publishing it would tell the people we are trying to stop exactly how to get around it. The rules themselves are public, and we would rather be asked about them than guessed at.

How to Report

Reporting from inside the app is the fastest route, because the context comes attached. On iOS and Android, press and hold the content and tap Flag. On web, click the ellipsis next to it and choose Report Message. Pick the reason that fits, and once the report is sent, the content is hidden from everyone in the room while we look at it. Characters, videos, rooms and profiles can be reported the same way, from the menu on each.

A trained moderator reads every report. Child safety and non-consensual intimate media take priority over everything else, and our moderators are available around the clock.

Some things have their own route. Copyright goes through our DMCA form. Misuse of your name, image, voice or likeness goes through our reporting form as a Likeness Concern. For anything else, our support form reaches the team directly.

Cantina Safety - photo of people's hands working at a table

Enforcement and Appeals

Enforcement is matched to the violation and to the account’s history. It can mean a warning, removal of the content, a temporary restriction on one feature, or suspension of the account, temporarily or permanently. Narrower actions exist so that a specific pattern of misuse can be addressed without suspending an account outright, and we tell you when one is applied.

If your account is suspended you will see a suspension screen the next time you sign in, with the option to contact Trust & Safety. You can also appeal through our Help Center. Every appeal is reviewed by a member of the Trust & Safety team against the specific policy at issue, and we tell you the outcome: the action may be reversed, upheld or modified.

When we suspend an account we also go back through what it created like its characters and the rooms it ran, and remove anything else that breaks the rules.