Analysing image…
Insights on image forensics, ELA analysis, metadata, deepfakes, and digital authenticity.
Anyone with a phone can now build a convincing fake of a face, and most of the raw material is already public. This guide walks through a two hour audit of every photo of you online: what a cloner actually needs, which images to remove, how to handle the ones other people posted, and how to keep proof of what is real.
Read article
Romance scammers rely on believable photos to build trust long before they ask for money. Learn how these fake identity images are made, the warning signs to watch for, and practical ways to verify photos before you become the next victim.
Read article
Glaze and Nightshade rewrite most of the pixels in an image, wipe its original metadata, and leave a noise signature that changes what a forensic scanner reports. None of that makes a fake pass as real. This guide explains what poisoning actually does to Error Level Analysis, metadata, and AI detection, and how to read a protected image correctly.
Read article
I signed a single JPEG, pushed it through Gmail, X, WhatsApp, Facebook and YouTube, then pulled every copy back and compared them field by field. One platform returned the file untouched. One quietly swapped my color profile for its own, and one deleted 29 fields before writing in a tag of its own.
Read article
The formats carrying most of the world's photos were built for compression efficiency, not for the forensic techniques we spent two decades developing on JPEG. This guide covers why Error Level Analysis goes quiet on HEIC and AVIF, where EXIF actually hides inside these containers, and how to convert a file without wiping the evidence you came for.
Read article
AI images stopped being obvious a while ago, and most community rules were written for a world where they still were. This guide covers the moderator's version of the problem: choosing a policy shape that survives scale, running a file check in under a minute, and handling the accusations that follow without losing half your team.
Read article