Acceptable Use Policy

1. Strictly Prohibited Activities & Content

You may not use the Zopyr AI gateway, nor permit any end user of your application to use the Services, to generate, distribute, or facilitate any of the following:

• Child Sexual Abuse Material (CSAM) or Child Sexual Exploitation and Abuse (CSAE) of any kind.
• Chemical, biological, radiological, or nuclear (CBRN) weapons development, or actionable instructions for violent extremism or self-harm.
• Offensive cyberweapons, zero-day exploit chains, ransomware payloads, command-and-control (C2) automation, or non-consensual credential harvesting.
• Automated disinformation campaigns, identity fraud, non-consensual intimate imagery, or phishing at scale.

2. Platform & Gateway Integrity

To maintain sub-25ms routing latency and high availability for all customers, you agree not to:

• Attempt to bypass RPM, TPM, concurrency, or free-tier rate limits by rotating fraudulent accounts ("Sybil attacks").
• Probe, scan, or test the vulnerability of Zopyr’s internal infrastructure or egress proxies without prior written authorization under our coordinated disclosure program.
• Forge HTTP headers, manipulate SSE billing settlement frames, or exploit payment gateway callbacks.

3. High-Stakes Automated Decision-Making Requirements

If your application operates in healthcare, legal services, consumer credit scoring, insurance underwriting, or critical infrastructure, you must implement qualified human-in-the-loop review and clearly disclose to end users that outputs are generated by artificial intelligence.

4. Violation Detection & Graduated Enforcement

Because Zopyr operates with zero-data-retention by default, abuse enforcement relies on automated rate-anomaly telemetry and upstream provider safety classifications. Severe violations result in immediate key revocation, while unintentional rate spikes trigger standard HTTP 429 Retry-After responses.