1. Respect rights and the law
Use the service only for lawful purposes and with the required rights to your inputs. Follow the applicable upstream model policies as well as these rules.
- Do not create or distribute child sexual abuse material, non-consensual intimate content, threats or material facilitating serious harm.
- Do not use the service for fraud, impersonation, credential theft, unauthorized surveillance, malware or attacks on systems.
- Do not infringe privacy, intellectual-property rights or confidentiality obligations.
2. Do not bypass controls
Do not bypass authentication, rate limits, billing, safety controls or account restrictions. Do not access another user’s data, distribute compromised keys, overload the service or conceal abusive traffic behind other users.
3. Keep humans responsible for consequential actions
Validate model output before using it for financial, medical, legal, employment or other consequential decisions. Your application must enforce permission checks and real-world constraints; an AI response is not authorization to perform an action.
4. Reports and proportionate action
Report suspected misuse to [email protected] with non-sensitive identifiers and a clear description. We may investigate, restrict access or take other proportionate action where needed to protect users, the service or legal obligations. You can contact us to explain or dispute a restriction.