Acceptable Use Policy

Approved by the owner on 2026-11-06; placeholders still to fill. Not legal advice. Before launch, check this policy against each enabled Provider's usage policy, since we must pass their limits on, and against the moderation categories in D-010.

This policy applies to everyone who uses [SERVICE]: account holders, organization members, and anyone using your API keys or your product built on ours. It forms part of the Terms of Service. Each model's Provider has its own usage policy, and you must follow that too; where the two differ, the stricter rule applies.

1. Content you may not create, upload or share

1. Child safety (zero tolerance). Any sexual or sexualized content involving minors, or content that looks like it does, including drawn, animated or fictional depictions. Also content that grooms, exploits or endangers children. We report it to the authorities and close the account.
2. Non-consensual and intimate imagery. Sexual or nude images or video of a real, identifiable person without their consent, including "undressing" or face-swapping someone into intimate content.
3. Real people. Images or video of a real person that falsely show them saying or doing something, made to deceive, harass or defame them, or to impersonate them, including public figures and election candidates. Satire and commentary that is clearly labelled is allowed where the Provider's policy permits.
4. Violence and harm. Content that promotes terrorism or violent extremism, depicts real-world graphic violence to glorify it, encourages self-harm or suicide, or gives meaningful help with weapons capable of mass harm.
5. Hate and harassment. Content that attacks or demeans people based on a protected attribute, or that threatens, bullies or doxxes anyone.
6. Intellectual property. Content that infringes copyright, trademarks or publicity rights, such as reproducing a specific artist's work, a film's characters or a brand's logo without permission.
7. Deception and fraud. Fake documents, IDs or receipts; scams and phishing; fake evidence; spam; and synthetic media passed off as real in order to mislead.
8. Illegal content. Anything illegal where you or your audience are, or that promotes illegal activity, such as selling drugs.

2. Things you may not do with the service

1. Try to get around moderation, rate limits or spending limits, for example by rewording blocked prompts on purpose, splitting requests, or opening extra accounts.
2. Probe, scan or attack our systems or other customers' data, or send traffic meant to disrupt the service. Report security issues to [SECURITY EMAIL] instead.
3. Share, sell or publish API keys, or resell raw access to [SERVICE] as a plain proxy without our written agreement. Building your own product on it is fine.
4. Use outputs to build a competing model where a Provider's terms forbid it.
5. Upload malware, or use reference URLs that point at private networks or at content you have no right to fetch.
6. Use the service from a sanctioned country or for a sanctioned person.
7. Use it to make decisions about people's eligibility for jobs, credit, housing, insurance or legal status.

3. If you build on [SERVICE]

1. Have your own terms that are at least as protective as this policy, and enforce them with your users.
2. Make it clear to people when they are seeing or hearing AI-generated media, especially realistic people, voices or news-like scenes, and follow any labelling laws that apply to you.
3. Don't remove watermarks or provenance data (such as C2PA or SynthID) where a Provider's terms or the law requires them to stay.
4. Pass on our safety signals: if we block a request, don't retry it through another route.
5. Give us an abuse contact, and help when we investigate abuse coming through your product.

4. Enforcement and reporting

1. Automated checks review prompts, uploaded images and outputs, and people may review flagged content. A blocked request isn't charged.
2. Depending on how serious and how repeated a breach is, we may block the request, warn you, lower your limits, revoke keys, or suspend or close the account. Suspected child sexual abuse material leads to immediate suspension and a report to the authorities.
3. If you think we got a decision wrong, reply to the notice or email [TRUST EMAIL] within [14] days. A person will review it. CSAM decisions can't be appealed.
4. To report abuse or content made with [SERVICE], email [ABUSE EMAIL] with the generation id or a link if you have one. Rights holders can send copyright notices to [COPYRIGHT AGENT / EMAIL].
5. We may update this policy as Providers and laws change, and we'll give notice as the Terms describe.