The ability to investigate a piece of media (where does it come from, when was it last edited, is its content real or fake...) is crucial in many domains. Today, we are releasing two new free online tools, alongside our AI detection tool, to help journalists, fact-checkers, trust & safety teams and fraud analysts.
Both tools run in your browser and require no account. Drag a file in, read the result:
Authenticity is rarely a single yes/no. The strongest investigations cross-reference several independent signals, and provenance metadata is one of the most valuable when present:
Used together with pixel-level detection, these signals help build a much more complete picture of a piece of media and spot contradictions. It could be a "photograph" whose metadata points to an image editor, or an image whose content credentials no longer validate.
C2PA (Coalition for Content Provenance and Authenticity) is a tamper-evident metadata standard adopted by OpenAI, Adobe, Google, Leica and others. When an image carries C2PA content credentials, they form a cryptographically signed record of how the file was created and edited.
Our online C2PA checker reads that record for you and shows:
The online C2PA checker reads and validates an image's content credentials. Click to enlarge.
This is the no-code companion to our Python C2PA tutorial: same underlying validation, but with nothing to install. Drag in an image and inspect its credentials in seconds.
One important caveat: C2PA data is easily lost. Screenshots, social-media re-uploads and most image-processing operations strip it. So the absence of credentials never proves an image is authentic or human-made. This is exactly why pixel-based detection remains essential as a complement.
Beyond provenance credentials, many image files contain embedded metadata. Our online EXIF metadata viewer surfaces all of it:
The online EXIF metadata viewer surfaces every metadata field a file carries. Click to enlarge.
For an investigator, these fields are a starting point for verification: do the timestamps line up with the claimed story? Does the device match the supposed source? Has the file been through editing software it shouldn't have? As with C2PA, metadata can be missing or deliberately altered, so it's a clue to corroborate rather than a verdict on its own.
No single signal is enough on its own. The most reliable approach is to combine them:
Whether you are a journalist reviewing sources, a fact-checker examining a viral image, a fraud or insurance analyst, or a Trust & Safety team, these tools give you fast, independent ways to check a file. For teams that need to do this at scale or without code, Sightengine Detect brings the same detection capabilities into a no-code dashboard, and the API integrates it all into your own workflows.
Try them now, they are free:
A developer's guide to C2PA, showing how to read metadata, detect tampering, and verify content authenticity with Python examples.
This is a guide to detecting, moderating and handling self-harm, self-injury and suicide-related topics in texts and images.