Sunday, November 12, 2017

Facebook: 'Send Nudes Through Messenger to Stop Revenge Porn'

by Tracey Smidt

In an attempt to stop revenge porn, Facebook has asked users to send nude pictures through its Messenger service instead of through other means across the platform. The images sent will be converted to a unique digital fingerprint which is then used to identify and block future uploads on Facebook, Messenger and Instagram.

Facebook has released a reporting tool in April allowing users to flag intimate photos that were posted without their consent to the community's operation team. Specifically trained representatives review the images to confirm if they are indeed in violation of Facebook's community policy before they are removed. After removal, photo-matching technology is used to prevent the images from being uploaded again.

This time around Facebook is a step ahead - the new technology lets users first upload images of concern to themselves on Messenger. From there, the images are converted to a unique digital fingerprint known as hash. Currently piloted in Australia, the company has partnered with the e-Safety commissioner's office.

So how would this work? Users are told to fill out an online form on the e-Safety commissioner's website, outlining their concerns. From there, users will be asked to send pictures they are concerned about to themselves on Messenger. Meanwhile, the e-Safety commissioner's office notifies Facebook of the submission. A community operations analyst from Facebook will access the image to manually confirm if it is in violation of policy. The image will be flagged as 'non-consensual intimate image' and be blurred out. This blurred image is stored for a short period of time and can only be accessed by a specially trained team before it is deleted.

Thereafter the image will be hashed. The hash will be retained and cannot be used to reconstruct the image. Flagged images cannot be uploaded again as they will be tested against the stored hash and blocked.

Photo matching technology was first developed by Microsoft in 2009. Working closely with Dartmouth and the National Center of Missing and Exploited children, the technology was used to put an end to continuous circulation of the same images of sexually abused children on the internet. Abusers got around this by altering files, either making a small mark on the image or changing the size.

Files can be recognized by the hashing system without direct access to it on the server. Actual content of an image is analysed and tagged, therefore bypassing attempts to alter files. As a result, altered images can still be identified, stopping distribution of images.

The hash technology is also used to take on other types of content such as child sex abuse and disturbing imagery.  Human reviewers will still be used to prevent legitimate images from being flagged as revenge porn.