ProvenanceGuard
← Go back

← All guides

What Happens to Provenance When Metadata Is Stripped From an Image?

By the Ingenious AI team · Published: November 14, 2026 · Reviewed: November 14, 2026 · 8 min read

The short answer

Stripping removes everything that lived inside the file: EXIF, XMP including a signed assertion, and any C2PA manifest. The pixels, and therefore the perceptual fingerprints and any visible label rendered into them, are untouched. So a stripped copy fails an exact-hash test, carries no embedded record, and still matches the registry through recovery, which is why the registry exists.

Seal an image free → Recover a reposted image Pricing

Think of the metadata as the luggage and the pixels as the passenger: stripping loses everything in the suitcase and nothing about who is travelling.

The pattern the simulations reveal is consistent and worth internalising. Embedded metadata, including the signed assertion and any C2PA manifest, does not survive the transforms. The visible label generally does, as long as it stays in frame and legible at the reduced size. The perceptual fingerprint still matches the registry record after the transform, within the match threshold, and the registry lookup therefore survives even when everything inside the file is gone. Distribution-proof provenance is the registry plus the label; the embedded assertion is the record of origin.

What is a stripped image? an image whose embedded metadata, EXIF, XMP and manifest data, has been removed, typically by re-encoding; its pixels are unchanged, and any provenance that lived only in the metadata is gone with it.

The problem behind metadata stripping and the layers

Running the check before a campaign, rather than after a takedown or a complaint, turns those findings into decisions. If the label fails at a platform's width, the artwork changes. If a format conversion in the pipeline strips everything before publication, the seal is applied after that conversion instead of before it. The simulations are cheap and repeatable, so they belong in the pre-publish routine for any asset that matters, not in the post-mortem.

Survival re-checks, a Pro feature, answer the question publishing actually raises: after this platform does what it does to uploads, what is left of my marking? The check runs local simulations of documented platform transforms on the sealed and labelled copies: resize to the platform's maximum width, JPEG re-encode at its quality setting, and metadata stripped. The result is reported per platform for Instagram, TikTok, YouTube thumbnails, X and LinkedIn, and every row is labelled as a simulation, because that is what it is, not an actual upload.

How ProvenanceGuard handles it

Two kinds of fingerprint do two kinds of work. The SHA-256 is exact: change one byte and it changes completely, which makes it perfect for proving a copy is bit-for-bit the registered file and useless for recognising a resized repost. The perceptual fingerprints, computed from the pixels rather than the bytes, are approximate on purpose: they stay close when a copy is resized, re-encoded or lightly edited, and drift as edits accumulate. A good registry stores both and is honest about which kind of match it is reporting.

The registry model also disciplines what a match means. A fingerprint match says this image is the registered image or a close descendant of it; it does not say the person showing it to you owns it, and it does not say the disclosure attached to it was correct. Those judgements stay human. What the registry removes is the argument about identity and timing: which file, registered when, carrying which declaration. Most real disputes shrink considerably once those three facts stop being contestable.

What survives distribution, and what does not

The registry is the part that survives distribution. At sealing, the asset's SHA-256, its sealed copy's SHA-256, and its perceptual fingerprints are stored against a public exhibit ID. Anyone with that ID, or with the file's hash, can open the public verification page or query the verification API and confirm that this exact file, or a near copy of it, was registered, when, and with what declared creator and disclosure. Verification pages are public by design: a record only a vendor can check would not settle disputes.

Labels have their own failure modes, and a serious workflow plans for them. A label placed along an edge can be cropped out; a label rendered too small becomes illegible after a platform resize; a screenshot of the centre of an image may exclude it entirely. That is why the visible label is one layer among several: the survival re-checks specifically test whether the label remains detectable after simulated platform transforms, and the registry fingerprint remains as the fallback when the label is gone.

A routine that survives busy weeks

There is also a tone question. A custody strip that looks like part of the work's presentation, clearly worded and consistently used, reads as professionalism rather than apology. Teams that label every AI asset the same way, from the first post rather than after a complaint, find the label becomes part of their brand's trust signal. The creators who struggle are usually the ones who labelled inconsistently, so that an unlabelled repost looks like concealment next to their labelled originals.

A visible label does the communicating that metadata cannot. Most people who see an image will never open its properties, and most platforms will never show them a credentials panel. A label in the pixels, naming the content as AI-generated, is read by exactly the audience the disclosure rules exist for, on exactly the surface where they encounter it. ProvenanceGuard's labelled copy carries that strip as part of the image, so the disclosure travels with the pixels instead of depending on a caption that can be separated from them.

Five actions to take

  1. Assume stripping will happen and design the record so it does not matter
  2. Keep the sealed copy yourself; the published copy is not your archive
  3. Download and keep both copies the seal produces: the sealed copy as the record, the labelled copy for publication
  4. File the exhibit's verification URL beside the asset wherever your team keeps campaign or client records
  5. Run check-a-copy on the version you actually published, and note which checks held: hash, signature, fingerprint

A worked example: An archivist receiving images back from a social platform export

Of 27 returned files, none carried its original metadata; using recovery lookups she reconnected 24 to their registry records, and the 3 that had also been heavily cropped were recorded as unrecoverable rather than guessed.

The lesson generalises beyond this case. Records made at publication time are cheap, calm and complete; records reconstructed under pressure are expensive, partial and contested. Everything in this guide is in service of moving work from the second category into the first.

FAQ

Can stripped metadata be restored into the file?

Not from the stripped copy itself; the data is gone. What recovery restores is the connection to the registry record, and re-sealing would create a new record for the stripped file, which is a different, later fact.

What formats and sizes are supported?

JPEG and PNG, up to 25 MB per file. Those are the formats the inspection, XMP embedding and fingerprinting are built and tested against; other formats are refused rather than approximated.

What does the free tier actually do?

The complete job on an asset: byte-level inspection, a signed assertion embedded as XMP, a visibly labelled copy, SHA-256 and perceptual fingerprints registered, and a public verification page. Paid plans add survival re-checks, evidence packs and bulk verification.

What do Pro and Team cost?

Pro is $29 a month and Team is $79 a month, as shown on the pricing page. Paid depth means per-platform survival re-checks, evidence pack export with EU, California and New York wording, and the bulk hash-verification API.

Seal an image free →