Document overview
This framework is designed to make AI disclosure, identity provenance, consent, knowledge boundaries, escalation and prohibited capabilities visible throughout the product lifecycle.
Legal review required. This structured draft supports counsel review and must not be treated as approved terms, a final privacy notice or formal legal advice until its publication status changes.
Governance principles
VowHumans experiences should be purpose-bound, proportionate, reviewable and designed around a named human or organisational owner.
Disclosure and human agency
People should know when they are interacting with an AI system, understand its role and retain a clear route to stop, leave or reach a human where the context requires it.
Identity and voice consent
Faces, voices and source media require documented authority, defined uses, expiry and revocation. Fictional or synthetic identities must not be represented as real people.
Prohibited capabilities
Product and customer controls should prevent deceptive or unsafe uses.
- Public-figure or unauthorised impersonation
- Appearance-based employment scoring
- Face-based emotion or honesty detection
- Hidden recording or undisclosed transcript capture
- Medical diagnosis or unsupported professional claims
Data and knowledge boundaries
Digital Colleagues should use approved, access-controlled knowledge; distinguish evidence from uncertainty; and avoid exposing private organisation or individual information.
Oversight, testing and incidents
Owners should test behaviour before publication, monitor outcomes, preserve relevant audit events, support abuse reporting and suspend unsafe experiences while incidents are reviewed.