Content loses its identity the moment it leaves your pipeline.

Diker preserves it. Origin, rights, AI production history, distribution path. Embedded in the file, recoverable even after metadata stripping, re-encoding, and social media upload.

Book a call How it works
Diker v1.0 IMG_2847.jpg
Content Credentials C2PA 2.2
Content fingerprint ISO 24138
Forensic watermark Imperceptible
Content matched
a7f2...c9e1 Original · Licensed

The problem

Metadata doesn't survive the second hop.

Origin

Wire service

WS_20260312_berlin.jpg

8/8 fields intact

Hop 1

Publisher CMS

berlin-summit-hero.jpg

5/8 rights lost

Hop 2

Syndication

img_28491.jpg

2/8 origin lost

Hop 3

Social media

photo.jpg

0/8 everything stripped

Creator, copyright, license, source, contact, credit, date, usage terms. Gone by the third hop.

How Diker works

Defense in depth.

Three independent layers, each embedded differently. Content Credentials in the metadata, fingerprinting and watermarking in the pixels. If one is stripped, the others still carry the full record.

Protected image

Content Credentials

Signed manifest in the file

L1

Content fingerprint

Computed from the pixels

L2

Forensic watermark

Invisible signal in the pixels

L3

Platforms strip
metadata

After distribution

Content Credentials

Metadata stripped

Gone

Content fingerprint

Match found in registry

Forensic watermark

Signal intact in pixels

Full record recovered.

Origin, rights holder, licensee, terms, integrity status. One lookup.

What this solves

Where did it come from. What was done to it. Who received it.

EU AI Act compliance

The answer is in the deliverable.

A regulator asks which AI tools were used. The declaration is embedded in the file. Every tool across the production chain, machine-readable and tamper-evident. August 2, 2026 enforcement.

Leak investigation

You know which copy and who received it.

Pre-launch content surfaces online. Every copy is uniquely marked. The watermark tells you which one leaked and where it was sent. Seconds, not weeks of auditing agencies and partners.

Institutional knowledge

Look at the asset, not a spreadsheet.

Someone needs to check usage terms three years from now. Talent rights, AI tools used, production history. It's all in the file, not in a spreadsheet someone has to find.

Who it's for

Built for the organizations that move licensed content at scale.

Wire services and photo agencies

You distribute thousands of images a day and lose track of most of them after the second hop. Diker gives every image a recoverable identity that survives metadata stripping.

Agencies and production companies

Your deliverables pass through four AI tools before mastering. Diker carries the full production declaration in the final file and checks it against the agreed tool stack.

Museums and cultural institutions

You digitize collections and license reproductions. Diker ties provenance, rights holder, and usage terms to every image file so the record travels with the asset.

Any organization publishing AI-assisted content in the EU

Article 50 of the EU AI Act requires machine-readable AI disclosure by August 2, 2026. Diker automates it in the production pipeline.

Automation with control

Handles the volume. Escalates the violations.

Fully automated for standard assets. When something needs judgment, the pipeline pauses and routes it to the right person.

You set the rules. The system handles the volume.

Integration

Content in, content out.

API-first. Send a file, get it back with provenance embedded. Batch processing, FTP watched folders, DAM integration. Minimal changes to existing tools.

Technical details
Book a call