Acceptable use policy

One policy for every creation

This policy applies to the website, workspace, API, submitted inputs and generated outputs, including requests made for your own end users. You must be at least 18 and follow applicable law, model conditions and the Terms of service.

Protect people and their consent

• Never create, request, upload or distribute sexual exploitation of children or sexualized depictions of someone under 18 or appearing under 18, including synthetic or fictional depictions.
• Do not create non-consensual intimate imagery, sexual blackmail or content that facilitates trafficking or exploitation.
• Pornographic or sexually explicit creations are prohibited. A model’s technical capability does not authorize a prohibited request.
• Obtain the necessary permission before cloning a real person’s voice, animating their likeness or placing them in a synthetic scene. Do not impersonate someone deceptively or fabricate their endorsement.
• Do not use creations to threaten, stalk, harass, expose private information or target a person or group with hateful abuse.

Avoid harm, fraud and illegal activity

• Do not provide operational assistance for violent attacks, terrorism, serious criminal activity or weapons intended to harm people.
• Do not encourage self-harm or provide instructions designed to enable it. Supportive and preventive discussion must remain safe and lawful.
• Do not create scams, phishing, fraudulent payment activity or misleading evidence intended to deceive people or authorities.
• Do not generate or deploy malware, steal credentials, exploit unauthorized access or automate abuse against people or services.
• Do not infringe copyright, trademark, privacy, publicity or other rights. Obtain rights for reference assets and respect model-specific commercial restrictions.
• Do not present AI output as qualified medical, legal or financial judgment, or use it as the sole basis for decisions that materially affect a person’s rights, health or livelihood.

Respect the platform and safeguards

Use documented interfaces and authorized access. Do not circumvent content safeguards, rate limits, account restrictions, quote validation or billing controls. Splitting a prohibited request into smaller steps or requesting another model to evade a restriction remains prohibited.

Lawful educational, artistic, journalistic and safety uses still need appropriate context and permissions. Context does not create an exception for child sexual exploitation, non-consensual intimate imagery or harmful instructions.

Enforcement and review

Requests can be refused and access can be limited when a violation or serious risk is identified. Responses may include removing access to content, warnings, feature restrictions, suspension or termination, depending on the circumstances. Legal reporting and preservation obligations apply where required.

If you believe an account action was mistaken, prepare a review request with the affected account or job identifier and relevant context. Do not repeat a blocked request to test ways around the restriction.

Report a concern

Use the content reporting page for harmful creations, unauthorized likeness or voice use, or rights violations. Include the content location or job identifier if available and a short factual description. Reporting options show which delivery channels are currently available. Never upload suspected child sexual abuse material as evidence.