Invisible Friends

Invisible Friends was considered ‘an ingenious way to put artificial intelligence to work for a good cause’, carrying out a task that humans simply aren’t capable of.

In 2017, we developed a world-first campaign to harness the power of Facebook’s then new facial recognition and auto tagging technology in the search for missing people. Invisible Friends featured a selection of Facebook profiles of missing loved ones whose families requested to be involved, sending through images and details to create the profiles. Through unprecedented media coverage both nationally and internationally (including coverage across Mashable and the BBC), we asked Facebook users around the world to ‘friend’ the loved ones involved.

The initiative searched the backgrounds of each photo and video posted by the friends of each profile, in the hopes of identifying and locating these missing people. With then around 500 million photos and videos posted to Facebook every day, if even one of the faces matched, the algorithm would have notified Missed of the missing loved one being identified in the image.

Facebook’s facial recognition technology operates at an accuracy of 98% which – at the time – was 13% more accurate than the facial recognition technology employed by the FBI. To allay privacy concerns, profiles were only created for loved ones where an active police report existed, and other strict criteria were met.

Though there weren’t any matches made, each of the families involved attracted their own localised media attention, and – through both social and traditional media – the issue reached a global audience.