Acceptable Use Policy

Don't use the Service to break the law, to generate CSAM or content sexualizing minors (zero tolerance, reported where required), to provide serious uplift for weapons capable of mass casualties, to conduct or facilitate cyberattacks, or to harass, defraud or intimidate people.

No credential sharing to evade per-user governance, no multiple accounts to dodge limits, no scraping the Site, no probing or disrupting the gateway, and no unapproved red teaming of Models (see the terms for the approval path).

Respect each Model's terms, including regional and organizational restrictions; circumvention with VPNs or proxies is a material breach. Where providers require use-case disclosures, answer honestly.

If you build on the Service, you're responsible for your users: appropriate human review for consequential decisions, disclosure that content is AI-generated where the law or common sense requires it, and rate limits or moderation suited to your audience. Burnable keys and per-key caps exist so that giving strangers access doesn't mean giving them your account.

Proportionate responses: warnings, key or model restrictions, breaker trips, suspension, and for serious violations termination with credits forfeited to the extent the law permits. We'll explain what happened unless the law prevents it.