Start with permission, not generation
A technically possible generation is not automatically safe to make or publish. If a real person can be recognised, confirm that you have clear permission for the specific use, audience and distribution channel before uploading their likeness or creating a synthetic performance.
Do not create deceptive identity content
- Do not impersonate a person to obtain money, access or trust.
- Do not place a person in sexual, violent, criminal or otherwise harmful fabricated scenes.
- Do not present synthetic speech or behaviour as an authentic recording.
- Do not use a public figure's likeness as an implied endorsement.
- Do not assume parody removes every legal, platform or safety obligation.
Record consent and scope
Keep a written record of who approved the use, what may be generated, where it may appear, how long permission lasts and whether paid advertising is included. For client work, make the approval part of the brief rather than relying on an informal message.
Disclose material synthetic media
Use clear disclosure where a reasonable viewer could mistake an AI-generated performance for a real event or statement. Also check the current rules of every distribution platform before publishing; labelling requirements and enforcement can change.
Review before export
- Confirm the depicted person and rights holder approved the final context.
- Look for unintended identity blending or resemblance to another person.
- Check captions, voice, logos and claims separately from visual quality.
- Store the brief, consent record, prompt and approved output together.
- Escalate uncertain commercial or legal uses to qualified counsel.
What this article is—and is not
This is a production-safety checklist, not legal advice and not a complete statement of afterSora policy. The current Terms, product controls and distribution-platform rules govern actual use.
