Technical

What Is C2PA? Tamper-Evident Photos, Explained Simply

A photo used to carry its own credibility. In 2026 it doesn't. Anyone can generate a convincing image of a flooded basement, a dented bumper, or a signed document in a few seconds — and anyone receiving a photo knows that.

EtchImage Field Guides · Updated August 2026 · 5 min read

C2PA is the industry's answer. It stands for the Coalition for Content Provenance and Authenticity, a standards group whose members include Adobe, Microsoft, Google, OpenAI, the BBC, and most major camera makers. The standard they publish is usually called Content Credentials when you meet it in the wild.

The idea in one sentence: attach a cryptographically signed record to a file that says where it came from and what has been done to it — so if anyone changes the image afterward, the signature no longer matches and the tampering becomes evident.

Check a photo's credentials in 5 minutes

  1. Get the original file, not a screenshot or a chat forward. Ask for the file as an email attachment or a direct download. Messaging apps and social feeds re-compress images and usually destroy the credential.
  2. Open verify.contentauthenticity.org (the Content Authenticity Initiative's free Verify tool) and drag the file in.
  3. Look for a signer, not just a green check. The useful question is who signed it — a camera, a phone app, an editing tool, an AI generator — and whether that signer is on the C2PA trust list.
  4. Read the timestamp. A credential countersigned by a trusted timestamping authority is far stronger than a claimed capture time, because it can't be backdated by changing a device clock.
  5. Read the edit history. Manifests can record actions like cropping, color adjustment, or "generated by AI." A crop is normal. An unexplained gap is not.
  6. Note what's missing. No credential doesn't mean "fake." Most photos on earth have none. It means you learn nothing either way.
  7. Keep the original. Archive the untouched file somewhere you don't edit. Once you export a copy, you may have exported the proof away.

What's actually inside a Content Credential

A C2PA manifest is a small block of data bound to the file, containing assertions (capture device, time, optionally location, edit actions, AI involvement) and a digital signature over a hash of the image itself.

That last part is what makes it tamper-evident. The signature covers the pixels. Change one of them and the hash no longer matches, so any validator will flag the file as modified. Nobody has to compare your photo to a "known good" copy — the math does it.

Edits don't have to break the chain. A conforming editor can record what it did and re-sign, producing a lineage: captured here, cropped there, exported by this tool. What you cannot do is silently swap the content and keep a valid signature from someone else's certificate.

Tamper-evident is not tamper-proof

This is the honest part most explainers skip.

Metadata gets stripped. Nearly every social platform re-encodes images on upload, and re-encoding produces a new file. Credentials frequently do not survive. Some platforms — LinkedIn and TikTok among the earlier adopters — read and display credentials, and Meta reads them to apply AI labels, but you should assume a credential dies somewhere between your camera and someone's feed unless you've tested that specific path. The industry's answer, "durable" credentials that pair the manifest with an invisible watermark and a content fingerprint so the record can be recovered from a stripped file, is real but still uneven in practice.

Screenshots erase everything. A screen capture is a brand-new image with no lineage. So is a photo of a screen.

Signing hardware can be attacked. In one publicized case, a camera maker suspended its authenticity service after a researcher demonstrated a way to make the camera sign an image it had not genuinely captured. A signature proves a key was used — not that reality was in front of the lens.

It says nothing about truth. A perfectly signed photo can still be staged, mislabeled, or taken three years before you're told it was.

C2PA raises the cost of forgery and makes casual tampering detectable. It does not make images self-proving.

Who supports it now

On the capture side, Leica shipped the first production camera with Content Credentials (the M11-P, in 2023), and Canon, Nikon, and Sony have brought signing to flagship bodies, some via firmware. On the software side, Adobe's Creative Cloud apps, and most major AI image generators, write credentials into their output. C2PA now runs a formal conformance program and trust list, so validators can distinguish an accountable signer from anyone who simply generated a certificate.

For phones, the practical route today is a purpose-built capture app rather than the stock camera.

Does it help legally or with an insurer?

Treat provenance as strong corroboration, not a magic pass. U.S. courts authenticate digital evidence under ordinary rules — Federal Rules of Evidence 901 and 902, including the provisions for certified records generated by an electronic process — and a signed manifest with a trusted timestamp is useful supporting material for exactly that showing. Insurers assessing damage photos face the same question adjusters have always faced, now harder: was this taken at the stated place and time, or pulled from a previous claim? Provenance makes that answerable. It doesn't replace an adjuster, a lawyer, or a forensic examiner, and no vendor should tell you otherwise. If a claim or case is significant, ask the professional handling it what format they want before you shoot.

Where EtchImage fits

The gap in all of this is the moment of capture. Most photos enter the world unsigned, and provenance added later only ever proves "unchanged since I uploaded it." EtchImage signs at capture: a C2PA credential with a verified timestamp, written before the file ever leaves the phone, so the chain starts at the shutter rather than at your inbox. It doesn't make a photo true — nothing does — but it means the version you hand over can be checked by anyone, with the same free tools described above, and that any edit after the fact will show.

The bottom line

C2PA is a signed receipt attached to an image. It's good at proving a file hasn't changed since a known signer handled it, decent at showing an edit history, weak against screenshots and social platforms, and silent on whether the scene was staged. Capture with it on when the photo might ever need to be believed. Keep the original. And when you're the one receiving a photo, check the credential — then keep asking the ordinary questions anyway.

Further reading: c2pa.org · contentauthenticity.org · C2PA technical specification · C2PA conformance program