Acceptable Use Policy

1. Scope

You are responsible for what you and your end users do with the Services. If your customers misuse the Services through your application, that is your responsibility to address. We expect you to have your own acceptable-use rules for your end users.

2. Prohibited content

You may not use the Services to generate, store, or distribute content that:

• Sexualizes minors, depicts non-consensual sexual content, or otherwise constitutes child sexual abuse material ("CSAM").
• Provides operational uplift for the development or deployment of biological, chemical, radiological, or nuclear weapons.
• Provides credible step-by-step guidance for committing physical violence against specific people, mass-casualty attacks, or critical-infrastructure attacks.
• Contains malware, exploit code targeting unpatched vulnerabilities for malicious use, or credential-stuffing tooling.
• Promotes or facilitates terrorism, human trafficking, or harassment of an identifiable individual.
• Defames, infringes intellectual property, or violates the privacy of an identifiable individual.

3. Prohibited conduct

• Reverse engineering, scraping, or attempting to derive model weights from outputs.
• Re-selling Perchy capacity as a generic API endpoint without our written permission.
• Probing, scanning, or testing the security of the Services without prior written authorization (we run a separate disclosure program — see /security).
• Using the Services to send unsolicited bulk communications or content that violates CAN-SPAM, CASL, or comparable laws.
• Circumventing rate limits, lane reservations, or billing controls (e.g., creating many free accounts to evade limits).
• Submitting personal data of children under 13, or sensitive categories of personal data (health, biometric, government ID), without a separate agreement that specifically allows it.

4. High-risk uses

Without a separately signed agreement that explicitly permits the use, you may not use the Services as a sole automated decision-maker in: medical diagnosis or treatment, legal sentencing, employment screening, credit underwriting, insurance underwriting, or life-safety systems. The Services are general-purpose tools; you must keep a qualified human in the loop for these decisions.

5. Third-party model rules

Each model provider has its own usage policy (e.g., Anthropic's Acceptable Use Policy, OpenAI's Usage Policies, Google's Generative AI Prohibited Use Policy). When you choose a model, you also agree to follow that provider's rules. We surface these references in the docs.

6. Reporting abuse

If you see content or behavior on the Services that violates this AUP, report it to abuse@perchy.ai. To report suspected CSAM, you may also contact NCMEC's CyberTipline. We treat reports confidentially and respond within one business day.

7. Enforcement

When we identify a violation we may, in proportion to the severity: warn the account owner, temporarily suspend access, terminate the account, refer the matter to law enforcement, or report to relevant authorities. We document each enforcement action and notify the account owner unless prohibited by law.