Acceptable Use Policy

1. General principle

Use the service for lawful purposes, respecting others’ rights, the Terms of Service and published restrictions of the selected model. Users are responsible for the legality of submitted material and their use of output. This policy does not prohibit good-faith discussion, research or defensive work within lawful limits.

2. Prohibited conduct

Fraud, phishing, theft of data or accounts, creation or distribution of malware to cause harm, unauthorised access and attacks on infrastructure are prohibited.

Child sexual exploitation and related material; organising violence or crime; unlawful trade in prohibited goods; harassment, threats, unlawful disclosure of personal information; and other content or conduct prohibited by applicable law are forbidden.

Do not impersonate others to deceive, infringe intellectual-property or other rights, send unlawful mass advertising, or use the service to evade legal prohibitions or security measures.

3. Technical restrictions

Do not bypass authentication, model permissions, load or balance limits; manipulate usage accounting; forge payment notifications; extract other users’ keys or data; or intentionally impose excessive load.

Keep API keys in protected application configuration and revoke them if compromised. Publishing a secret key in a public repository or sharing it with unknown persons creates a risk of unauthorised spending.

4. Sensitive data and consequential decisions

Do not submit bank codes, passwords, active secret keys or information you are not entitled to transmit. Special categories of personal data, medical confidentiality, biometrics and other specially protected information require an established legal basis and confirmed processing conditions before use.

Verify model output before actions that materially affect health, safety, money or people’s rights. The service does not replace required professional review, and a model’s suggestion does not make an unlawful decision permissible.

5. Investigation and measures

Technical metadata, abuse reports and information voluntarily supplied to support may be reviewed to protect the service. This does not mean publishing user requests or giving other users access to them.

Depending on the circumstances, measures may include a warning, request-rate restriction or suspension of a key, feature or account. An immediate threat may require immediate restriction; otherwise severity, repetition and the possibility of correction are considered.

Where consistent with security and legal obligations, users receive the reason and an appeal route. Unused paid value is handled under the Refund Policy; suspension is not an automatic penalty or forfeiture of the balance.

6. Reporting and appeals

Report abuse, compromised keys or vulnerabilities through account support or support@tokeniser.ru. Include a safe description, time and transaction identifiers; do not attach active secrets or other people’s personal information. Test vulnerabilities only on systems you are authorised to test.

An appeal should identify the account, circumstances and explanation. It is reviewed using available technical evidence; statutory remedies remain available.