Recognised Faces is an internet application that generates a daily image of a face from images found via google’s lists of top search terms. Facial features in the found images are identified, using facial recognition technologies usually reserved for mass surveillance, before being combined into an image of a new face. After being generated these faces are used as the personal avatar of Kristoffer Ørum on his website, on various social networks and anywhere else his image might be indexed and scanned for facial features by intelligence agencies, commercial agents or other interested parties. By constructing new faces from parts of the most looked upon images on the internet Recognised Faces creates a snapshot of the flow of data collection and facial recognition that happens daily on the internet, thus utilising facial recognition to generate phantom faces that reflect how computers perceive us as vaguely recognisable patterns in an ocean of data. When these phantom images are fed back into the internet, they may help to destabilise the NSA’s or Google’s images of who Kristoffer Ørum is ever so slightly. The glitchy faces that emerge from the computer’s dispassionate gaze clearly differ from how faces appear to a more human gaze. They may appear somewhat monstrous and weird, but for the most part they remain strangely reminiscent of the beauty ideals that dominate mainstream media as well as most of the internet. What to human eyes might appear to be errors and distortions reveals traces of the statistical mode of perception that is really at work here - illustrating shortcomings of much reviled surveillance technology while providing us with a mechanical mode of observation that just might reveal things about our species that our own perception is unable to show us.