C2PA: From Theory to Practice
TL;DR
- C2PA works, but production needs resilience, trusted identities, clear summaries, and safe redactions.
- We turn long manifests into a simple traffic-light view you can act on.
- Here’s how we’re shipping this at Wolftech—and what publishers can do next.
Lessons From History
After World War I, France built the famous Maginot Line—a massive wall of bunkers, tunnels, and artillery to stop another invasion.
It was impressive, but it failed. In 1940, the German army simply went around it.
The lesson? Defenses that can’t adapt will break.
C2PA—the open standard for signing and tracing media—is our modern “line of defense.” But if we treat it like a rigid wall, it could end up being another Maginot Line. To succeed, it has to bend, flex, and fit into the real workflows of newsrooms.
At Wolftech, we’ve taken C2PA from demos into production. Here’s what we’ve learned about making it work under real-world pressure.
From Lab to Newsroom
We’ve taken C2PA out of the lab and into newsroom workflows step by step. In our first implementation, editors can instantly see whether a file has a manifest, who last signed it and when, and—if a photo or video is made from multiple parts—an overview of the sources.
We’re now adding a timeline of changes. That means you don’t just see the current state; you see how the media came to be. This poses a new challenge, because manifests can be long. A “simple” image might span 14 pages, and a video built from a dozen clips can have a manifest of close to 100 pages. That detail is a feature, not a bug: it captures every edit. But nobody reads 14-100 pages on deadline. Our job is to turn that richness into quick, usable clarity.
The Pillars of Production
Moving to production changes the question from “Does it work?” to “Does it deliver value on deadline?” For us, four areas matter most.
1. Resilience across the supply chain
Not every newsroom tool supports C2PA, and that won’t change overnight. Provenance shouldn’t disappear when a file is resized, transcoded, or passed through legacy software.
Why it matters
The chain is only as strong as its weakest link. If provenance breaks anywhere, trust suffers. Journalists spend extra time re-verifying images, and audiences may lose confidence when content credentials aren’t present.
What we do now
Long term, every tool should support C2PA end to end. Short term, we keep provenance alive with complementary methods:
- Detached sidecars that carry the manifest alongside the asset
- Content fingerprints or watermarking to bind a file to its manifest
No single method is perfect, but together they survive real workflows. We have to tread carefully, though, to avoid exposing redacted details. We’ll cover confidentiality and safety later.
2. Identity and trusted signing (Device vs. Organization)
Early C2PA led to a common question: who signed this—camera, journalist, or publisher? The ecosystem now has a clear path with the CAWG Identity Assertion, supported by IPTC trust lists.
What it unlocks
Cameras can sign at capture. Publishers can add an organization certificate later. Verifiers can display a name editors recognize (e.g. “Signed by BBC/AFP/CBC”)—not just a certificate string. That turns signatures into human signals, not just machine checks.
3. Usability: from data to insight
C2PA captures everything, but editors need only the right things, fast. With limited screen space, time, and attention, turning long manifests into quick decisions is a challenge the whole broadcasting community must solve for C2PA to succeed.
Our approach
Replace undifferentiated manifest dumps with rule-based summaries that group edits by impact:
- Harmless: resizing, transcoding, color-profile normalization.
- Context-affecting: cropping, cutting, time-compression.
- Transformative / high-risk: heavy compositing, content-aware fills, AI manipulation.
Once grouped, we will present a compact digest with:
- A simple, explainable “nutrient” score (a quick read on edit impact and provenance completeness), and
- A health check, covering manifest completeness and the presence of history edits.
The result is a traffic-light overview editors can act on at a glance: green for harmless, amber for context-affecting, red for high-risk.
4. Confidentiality for journalists
Provenance is powerful, but it can expose sensitive details like a photographer’s identity, GPS location, or device IDs. Sometimes that data must be redacted to protect people. If we remove it silently, users may assume something’s wrong.
Our principle: transparent redaction.
Every removal comes with a short note that says what was removed and why. For example, “GPS removed to protect a source.” This balances safety with respect for the audience—no surprises, no gaslighting.
Edge cases we account for
- Linked versions: If earlier versions can be reached via watermarks or digital fingerprints, ensure they don’t re-expose redacted faces or locations.
- Previews and thumbnails: Apply redactions consistently so thumbnails and provenance previews don’t leak sensitive details.
These are the kinds of details we need to solve carefully to make C2PA truly production-ready.
A Practical Checklist for Publishers
If you’re ready to move from theory to practice, here’s a starting point:
- Get an organizational certificate aligned with CAWG.
- Adopt tools/validators that support CAWG + IPTC trust lists.
- Preserve manifests via sidecars/remote storage; add fingerprints and watermarking for resilience.
- Define edit categories (harmless / context‑affecting / transformative) and publish them internally.
- Implement redaction notes (“what and why”) for safety‑driven removals.
- Ship a digest UI: grouped changes + nutrient score + full drill‑down.
- Train for speed: integrate the digest right where editors make decisions.
Not Another Maginot Line
The Maginot Line was brilliant, but brittle. C2PA can be different because it’s adaptable, and because the community is aligning on identity, resilience, and usability. The standard captures the facts. Our job is to surface the meaning, fast—without putting people at risk.
Production‑ready isn’t the finish line—it’s the starting grid. Now let’s make provenance visible, usable, and safe for the people who need it most.

Sergej Stoppel
Sergej Stoppel is the chief innovation officer at Wolftech, an Avid brand, and leads the development of a collaborative news and media production platform used by over 20,000 professionals worldwide each day. He holds a PhD in Computer Science (Visualization) from the University of Bergen, where his dissertation was recognized with the EuroVis Best PhD Award in 2019. His work has received multiple international honors, and he is widely regarded as a thought leader in data science, deep learning, and human-computer interaction.