Detecting AI-augmented media forgery (w/ WAT.AI).

Today's media forgery tools are often strong at telling apart fully AI-generated media from fully human-made media. They struggle when the truth sits in between: real footage or photos that have been partly altered by AI. That middle ground is where modern OSINT work is most exposed.

In partnership with WAT.AI, we are building a detection tool aimed at that gap. The goal is to confidently flag AI augmentations in images and video, not only the easy 100% AI or 100% human cases.

We are building it for law enforcement analyst units that need to check whether media used in an investigation was partly altered.