Trust Engine

Recover source first. Explain uncertainty second.

ContentSeal is designed for the moment after metadata is lost. It separates exact file proof, screenshot/repost recovery, visible-content differences, issuer accountability, and lifecycle status.

01

Media intake

Accept an upload or safe direct image URL, validate image type and size, and generate a media ID.

02

Exact source check

Compute SHA-256 to detect whether the scanned image is the original sealed file.

03

Visual recovery

Use perceptual hashes and crop-aware comparison to recover a source from screenshots, crops, compression, or reposts.

04

AI evidence layer

Explain source recovery, human accountability, synthetic-media context, and change risk without pretending it is a truth detector.

05

Difference signals

Check metadata loss, OCR conflict, C2PA, watermark, classifier, status, and version signals separately.

06

Content Trust Card

Generate a human-readable summary with verified facts, uncertainty, and recommended action.

Trust labels

Clear labels, cautious claims.

Verified Original

Exact file hash matches an active proof receipt.

Changed Copy Risk

A source was recovered visually, but the uploaded file differs from the sealed original.

Source Recovered From Screenshot

A source was recovered after screenshot, crop, repost, compression, or metadata loss.

Expired Content

The receipt was valid before, but should not be treated as current.

Older Verified Version

A known receipt exists, but a newer version should be checked.

No Verified Origin Found

No matching proof receipt was found. This does not prove the content is fake.

Conflicting Signals

Some signals match while others conflict with the proof receipt.

Try the engine

Scan an upload or paste a direct image URL through the recovery console.