Content provenance

What it is

A provenance record is a tamper evident manifest attached to a file and signed by a certificate, listing details such as the capture device, the software used, the edits applied and whether generative AI was involved. It complements older embedded metadata such as EXIF and IPTC fields for creator, credit and licence, and on page markup that names the creator and rights holder. Because the manifest is signed, changes made without updating it can be detected.

Why it matters

As synthetic media becomes cheap to produce, provenance is one of the few durable signals of authenticity, and platforms use it to decide what to label as AI generated. For publishers and stock libraries it is also an attribution mechanism: creator and licence data that survives republishing keeps credit and traceability intact. In high stakes topics such as news, health and finance, verifiable media supports the wider trust signals that search and AI systems look for.

How it works

Practitioners switch on Content Credentials in supported cameras and editing tools, then audit their publishing pipeline so resizing, CDN transforms and CMS uploads do not strip metadata. They fill IPTC creator, credit line and licence fields consistently, mirror the key facts in on page markup such as schema.org creator and copyrightHolder, and mark AI generated visuals both in metadata and in the visible caption. Rights teams use the same fields to find unauthorised reuse.

When it applies

It applies to any organisation publishing or licensing photography, video, audio or AI generated visuals, and is most valuable where authenticity and attribution carry commercial or editorial weight.

Examples

• A newsroom preserves Content Credentials on press photographs so readers can inspect capture and edit history.
• A stock library keeps IPTC creator and licence fields through its resizing pipeline so credit survives republishing.
• A brand marks AI generated product visuals in the file metadata and states it in the on page caption.

How it is measured

• Share of published images and videos that retain provenance or IPTC metadata after processing.
• Proportion of media assets carrying a valid signed manifest at point of publication.
• Number of unauthorised reuses identified through creator metadata or reverse image search.
• Rate of correct attribution when third parties republish your media.