• Terms of Service
• Privacy Policy
• Acceptable Use Policy
• Cookie Policy
• Content & Safety Policy

Content & Safety Policy

Last updated 6 September 2026. Version 1.0. Part of the Terms of Service; the prohibited categories are in the Acceptable Use Policy.

How safety works on Nixra

1. Safety filtering is always on. Every generation model runs with its safety checks enabled at the safest setting the model's provider supports. This is fixed on the server, per model.
2. There is no setting to weaken it. There is no "NSFW mode", no relaxed option, and no user or client input reaches any safety parameter. Requests that try to inject a safety flag are ignored or rejected by the server.
3. Layered enforcement. The providers that run our models (fal.ai, kie.ai and Runway) apply their own moderation on top of ours; content can be refused by either layer. The always-on rule applies identically to every provider we route to.
4. Uploads are verified server-side. Uploaded media is checked at the byte level: file type sniffing, size and duration measurement. Declared metadata is never trusted.

What this means for you

• A generation that trips a safety system fails safely: the job is rejected or fails and the reserved credits are refunded automatically.
• Repeated deliberate attempts to produce prohibited content are an Acceptable Use breach and are handled under its enforcement section.

Real people

Content depicting a real, identifiable person carries specific rules. Intimate imagery without consent is always prohibited and, in the UK, a criminal offence to create or to request. Deceptive impersonation is prohibited. Clearly labelled satire and parody of public figures is permitted. When you upload a photograph of a person you confirm you have their permission. If you publish an AI-generated depiction of a real person, make clear it is AI-generated wherever the law where you publish requires it.

Reporting content

Use the report control on any of your own results. For anything else, email abuse@nixra.ai with the URL. We review within two business days, within 48 hours for intimate imagery of a real person, and immediately for anything involving minors.

Transparency about limits

No automated safety system is perfect. Filters occasionally block harmless requests and occasionally miss harmful ones. The reporting channel and the enforcement process exist to close that gap. We do not claim more than the system does.