Facebook May Soon Identify and Track Users through The Photos They Upload
Facebook’s facial recognition technology is being perceived as a privacy threat but it seems the social networking giant is not done with its privacy-offending developments yet. Just recently, Geek.com published a story about their discovery of a patent filed by Facebook over a technology that seeks to “fingerprint” images to identify and track their owners. It’s just a patent for now but the idea can certainly send chills to every privacy-advocate’s spine.
The Patent
Facebook recently applied for a patent to associate cameras with user and objects in a social networking system. This was filed on January 14 this year and published on May 7 with publication number US20150124107 A1 and application number US 14/596,483. The named inventors are Daniel Gregory Muriello, Stephen Charles Heise, and Jie Chen and the original assignee is Facebook, Inc.
Basically, the patent is about a “system for associating cameras with users and objects in a social networking system.” It involves the analysis of photos uploaded to a social networking system to find signatures of the cameras used to take the photos. These signatures will then be used to establish associations between the cameras and their respective social network users.
Camera Fingerprinting
The EXIF data of images can reveal a good amount of information about the photos being uploaded online. However, these data can be lost when the images are edited. Also, they can identify the devices and camera settings when taking photos but they cannot specifically pinpoint which specific device took the photo. It is limited to naming the model number or technical name of a device.
To specifically identify a device and subsequently the owner of the device, a form of “camera fingerprinting” is needed. This fingerprint is not found in the EXIF data but in tiny distinctive attributes that are distinct and unique to devices. They include the following:
- image artifacts
- faulty pixels
- scratches on the lens
- color bleed
- lens-related discolorations
- lens-related distortions
Regardless of the brand or model, cameras may have distinctive attributes that manifest in the photos they produce. Image artifacts can be software or hardware related and can be distinct to a particular device model. Faulty pixels is an issue that can be unique to specific devices, similar to defects on images resulting from scratches on the lens. Color bleeding could be an image processing issue that could result from the sensor or from the image processing software used. Discolorations and distortions associated with the lens used on a device may also help distinguish a device from others. Individually, these image attributes cannot distinguish a photo-taking device from others. However, when they are taken into account together, the can help form the “camera fingerprint” Facebook is looking at.
These don’t compare to the uniqueness of a fingerprint but at the very least, they are one of the few more reliable ways of identifying an individual on social media, especially when used in combination with other information like file naming conventions, location data, and Facebook’s own facial recognition tech.
Purpose of the Technology
As mentioned, associations will be established between a social media user and the camera he or she uses. These associations will then be used for various purposes including the determination of what events or groups to recommend, the identification of multiple user accounts created by the same person, finding out if an account is fraudulent, and knowing the possible relationships or affinity between social network users. Facebook may also use this system to track down users who create multiple accounts, something that is supposed to be prohibited by Facebook’s system.
In short, this is an attempt to make use of photos in tracking social media users in the guise of improving features, functions, or services. Facebook, to date, already hosts over 250 billion photos on its servers. It is estimated that Facebook users upload around 350 million photos every day into their accounts. Apparently, the people behind the social media giant want to make the most of what they can do with the huge volumes of photos they have to keep in their servers.
Public Reaction
Based on blogs and posts on social media, it’s safe to say that not many are happy with the idea Facebook is pursuing. Some would even call it creepy. Facebook claims that this “image fingerprinting” system is supposedly meant to enhance its services like the way it makes recommendations for friends and determining affinity among users. However, it is not being too paranoid to think that it can violate user privacy. It’s far worse than the cookies and other methods used by advertisers and website owners.
Not Inescapable
The good thing, however, is that this potentially-privacy-violating system patented by Facebook is easy to defeat. It is highly likely that new image-editing services or apps will be developed to counter this image fingerprinting scheme. It will be easy to randomize the defects or to make random modifications on an image to prevent Facebook from creating associations between images and their respective owners.