This policy forms part of the Terms of Service. It covers everything you send through the Selora API, everything you build on it, and anything directed at the platform itself. Upstream model providers may apply their own restrictions on top of this policy, and using a model means accepting those too.
1. Illegal activity
Do not use Selora for fraud, scams, phishing, identity theft, money laundering, sanctions evasion, or trafficking in people, drugs, weapons, or stolen goods, and do not use it to infringe copyright, trademarks, patents, or trade secrets.
2. Harm to people
Do not use Selora to create or distribute child sexual abuse material or sexual exploitation of any kind — such accounts are terminated without warning and reported to the authorities. Do not use it for instructions for mass-casualty weapons, encouragement of suicide, self-harm, or eating disorders, harassment, stalking, doxxing, or incitement to violence or hatred against protected groups.
3. Malicious code & unauthorized access
Do not use Selora to build malware, ransomware, spyware, or botnets; to develop exploits for systems you are not authorized to test; to credential-stuff or defeat CAPTCHAs; or to hide intrusions. Security research on your own systems, authorized penetration testing, CTFs, and education are fine.
4. Deception & impersonation
Do not use Selora to impersonate real people or organizations, to make deceptive deepfakes, to run disinformation campaigns or manipulate elections, to fabricate reviews or metrics, or to present AI output as human-authored where that would materially mislead someone.
5. Spam
Do not use Selora for unsolicited bulk email, SMS, or messaging; doorway pages; mass low-value SEO content; or engagement farming.
6. Privacy & surveillance
Do not use Selora for unlawful processing of personal data, mass surveillance, tracking people without consent, facial or biometric identification of private individuals, inferring protected traits in order to discriminate, or large-scale scraping that breaches a site's terms.
7. High-stakes decisions
Model output can be wrong. Do not present it as professional advice, and do not let it decide medical, legal, credit, insurance, employment, housing, or immigration matters, or safety-critical control, without qualified human review.
8. Platform abuse
Do not:
- Bypass or evade rate limits, token accounting, rolling windows, or billing — including by registering multiple accounts.
- Create extra accounts to obtain more free-trial allowance than one person is entitled to.
- Share, resell, or redistribute access to the gateway without a written agreement with us, or build a directly competing gateway on top of it.
- Attempt to extract, reproduce, or reverse-engineer model weights.
- Probe, scan, or stress-test our infrastructure without written permission.
- Send deliberately malformed requests to disrupt or confuse the gateway.
9. How we enforce
We do not read or store the content of your prompts, so enforcement relies on usage patterns, abuse reports, notices from upstream providers, and payment signals. Responses range from asking you for an explanation, to throttling a key, suspending an account, and termination with balance forfeiture as described in the Terms, up to referral to law enforcement. Reports of child sexual abuse material or credible threats to life are acted on immediately.
10. Reporting a violation
Report abuse — including abuse of someone else's account or keys — by emailing [email protected] or through the contact page. Include enough detail to locate the account or traffic concerned.
11. Appeals
If your account or key is restricted and you believe that is a mistake, email [email protected] or send a message through the contact page. Every appeal gets human review, and we will tell you the outcome and what would be needed to restore access.