Joshua Krinsky

I work on deep image fake creation and detection and, increasingly, on the human side of detection: which manipulations actually mislead people or cause harm, what information helps them judge an image, and how to design interventions that turn a detection signal into something an end user can act on.

A detector that is accurate in a benchmark and unusable in an interface and understanding impact has not solved the problem it was built for. My interest is in where that gap actually opens up, and what design can close it.