Responsible AI Principles
Before using an identifiable face or voice, establish permission, presentation and intended audience. Consider whether the result implies an event or endorsement that never occurred. Clear attribution and rights handling are part of publication quality.
Ask about consent and context
Before using an identifiable face or voice, establish permission, presentation and intended audience. Consider whether the result implies an event or endorsement that never occurred. Clear attribution and rights handling are part of publication quality.
For historical and religious imagery, distinguish documentary sources from symbolic reconstruction. Identify constructed scenes where necessary and inspect assumptions in appearance, clothing and roles. Keep sources and revision notes so an error can be corrected.
Review what an image implies
A generated presenter holding a branded product can imply endorsement even when no written claim appears. Review the relationship between identity, object and caption together. A disclosure hidden elsewhere may not resolve an implication made by the image itself.
Create an asset note
For a proposed scene, write who or what is recognizable, the source reference, the intended use and any permission still needed. Label reconstructions beside the image. If permission or provenance is uncertain, choose a documented alternative before publication.
