Trust and safety
Voice & Synthetic Media Policy
Consent, evidence, prohibited impersonation and withdrawal rules for voice cloning.
Allowed rights bases
You may create a clone only when it is your own voice or you hold current, specific and written authority from the voice owner. The authority must cover AI voice cloning, intended uses, disclosure, providers and retention. Keep the original evidence. Pynokio may request it and may suspend the clone while reviewing a complaint.
Not allowed
- Cloning a person without authority, including a public figure, colleague, customer, family member or deceased person whose rights remain protected.
- Fraud, social engineering, bypassing voice authentication, fake evidence, political manipulation, deceptive endorsements or emergency impersonation.
- Sexual, humiliating, threatening or discriminatory uses; targeted harassment; or evasion of another platform's safeguards.
- A minor's voice through the standard self-service flow. Verifiable guardian authority and enhanced operator review are required.
- Using a clone after consent/authority has expired or been withdrawn.
What Pynokio records
The workflow records a versioned statement, declared rights basis, optional written-authorisation reference, account identifier, hashed network identifier, user agent, timestamp and source job. This supports accountability but does not by itself prove that the declaration is true. Samples are private and are not served as public files.
Disclosure and provenance
Generated speech carries an inaudible Pynokio audio watermark and a machine-readable provenance record. Anyone can check a file with the free Audio Detector. You must also make a clear audience-facing disclosure when the output imitates a real person or could be mistaken for an authentic recording. Do not remove, alter, obscure or circumvent the watermark or provenance.
Withdrawal, deletion and complaints
The account holder can delete a voice; this revokes its consent record and removes operational samples/preview files. The voice owner may contact [email protected]. Include identity/contact details, the voice or asset reference, why you own/control the rights and the requested action. Pynokio will preserve only the minimum evidence needed for investigation, legal obligations and claims.
Special-category and sensitive information
Voice can reveal identity, health, origin, beliefs or other sensitive information. Voice data becomes biometric special-category data when technically processed to uniquely identify a person. Pynokio's generation workflow must not be repurposed for speaker identification or biometric surveillance.