Why likeness needs its own consent question
Generative tools can transform an ordinary reference into media that depicts events, bodies, speech, or contexts that never existed. The relevant question is therefore not only whether someone supplied an image, but whether they agreed to this type of transformation and its audience. Permission should be specific enough to cover purpose, medium, distribution, and material changes in context.
This matters across face swap, voice synthesis, reference-image generation, lip sync, and character-consistency systems. A tool's technical ability to reproduce an identity does not grant the right to use it.
Consent is not the same as access or resemblance
- Public availability is not permission. A photograph posted online can still depict a person who never agreed to synthetic reuse.
- Possession is not authorization. Receiving an image privately does not authorize transformation or publication.
- Similarity is contextual. An original synthetic character may incidentally resemble someone, while deliberate use of an identifiable person's name and reference images raises a much clearer likeness issue.
- Consent can have boundaries. Permission for a private draft, a non-adult portrait, or one platform does not automatically extend to other uses.
Adult-AI use
Non-consensual intimate synthetic media can cause direct harm to an identifiable person and may violate criminal, civil, platform, or distribution rules. This glossary explains the concept; it does not provide methods for creating or evading safeguards around such material.
Adult-AI services including PornCreate, Pornworks, and other reference-driven tools differ in how they state real-person rules, verify age, moderate uploads, or retain data. Users should treat documented restrictions as a floor, not as evidence that every possible upload is consensual.
Consent, verification, and provenance
Age verification addresses whether a user or depicted participant meets an age requirement; it does not by itself prove permission for a particular generation. Face or identity verification can help link an account to a person, but still may not establish the scope of permission.
Content Credentials and watermarking can communicate origin or editing history in some workflows. They are not consent records unless a broader system deliberately binds an authorization to the asset, and even then the authorization's scope must remain understandable.
Practical consent principles
For identity-linked synthetic media, a responsible workflow uses original fictional identities, the creator's own likeness, or people who have clearly authorized the specific use. It avoids using ambiguous, scraped, or third-party images as proof of permission; documents the intended audience and distribution; and provides a way to stop further use or distribution where feasible.
Laws differ by jurisdiction and are changing, so platform policy and this general explanation are not substitutes for legal advice. The stable principle is simpler: the more intimate, realistic, identifiable, or widely distributed the output, the less defensible it is to rely on implied permission.
Common misconceptions
- A disclaimer that media is AI-generated does not cure a lack of consent.
- A fictional label does not prevent identification when the output clearly uses a real person's likeness.
- Automated moderation cannot guarantee that all uploads are authorized.
- Consent to model, photograph, or date someone is not blanket consent for synthetic intimate media.