Consider this scenario: you are at a party with your friends and someone decides to snag a selfie with you. Later on they are tagging you in the photo as well as noting this happened at so-and-so’s house. Now, while all of your friends maybe interested in seeing the fun times you had at this party, you would be surprised that a number of different businesses and organizations would also be interested in these photos of you and your friends. In fact they are very, very, very interested when and where you and your friends go, what you do, buy, eat, and even how you get there.
All of this is prized data in today’s world and those of us with internet capable devices are walking, talking sources of data on how to better sell products. However, some are concerned that this constant state of meta-data gathering may be crossing a line into the privacy of people’s lives. Saying that some methods of data retrieval lack proper transparency for individuals to properly understand and therefore collect more information than people originally intended to give.
One such example of this is the program DeepFace, developed by social media giant Facebook. DeepFace improves upon the current facial recognition software that has been out for decades by deep learning where in DeepFace is tasked with one task, distinguishing between faces, and through a process of trial-and-error learns to become better at the task. By being a computer program, DeepFace has the advantage of computer processing speeds which allows it to perform the task dozens and dozens of times a second, and learn at a rate much quicker than a human. This is the reason why DeepFace can now successfully identify individual’s faces in photos with 97.35% accuracy, and humans averaging ~80% accuracy when performing the same task.
Facebook has said that it wants to use this technology to help automatically identify users when photos of them are uploaded by one of its 1.3 billion users, and Facebook does have a lot of photos uploaded on a daily basis. With an estimated 400 million photographs uploaded on to the site every day. Ideally, Facebook says that users would receive an email notifying them about the photograph and giving them the option to blur their face out of the photograph. This would help individuals keep compromising photos of themselves from reaching unwanted eyes. While this does sound appealing in concept it does raise concerns regarding the storage of data and the consent of the photo’s subject.
Facebook keeps these photos stored in their Social Face Classification (SFC) system which is used to help train DeepFace towards this goal, and this storage makes some people nervous about the amount of data that Facebook is storing on them. The thought of every smart phone and camera now having the potential to recognize individuals, even in a crowd, is of concern to them. As for individuals it would mean the end of anonymity as now whenever they are out of their house they are now potentially being filmed, recorded, and archived in some form of database. For others it is the idea of this being used for targeted marketing as businesses could use the facial recognition software in DeepFace, or a similar product, to keep more detailed records on individual customer behavior. Using this software to recognize customers in their stores and being able recommend items to them based on previous purchases and other meta-data gathered on them. Lastly, some are concerned about the consent of non-Facebook users who may now have data about them stored in DeepFace’s SFC system. This could range from individuals who have intentionally chosen to not have a Facebook account to children who may not have a Facebook account.
Apart from DeepFace, smartphones are also sources of meta-data on consumer behavior. ‘Geo-tagging’ has become a popular way for businesses to track consumer travel patterns and better target their advertisements based off of that data. This can include what restaurants the individual passes on their morning and evening commutes, deals on clothes near local shops they frequent, or even advertisements on apartments when the individual’s lease is nearing its end. While some of this may sound appealing to the bargain hunter in all of us, it also ranges issues of security and privacy due to the method in which this data is gathered. By tagging yourself to a location, this also allows lets others know where you have been. Most mobile devices can remember the networks that they have connected to in the past. It is how your cellphone can automatically log in to a friend’s wifi network after the first time and not requiring you to manually log in each time after. This data can be collected and analyzed to show what previous networks individuals have logged into and use this to get a geographical sense of where they have been.
So what does all of this mean? For many this is fast becoming the new normal. That we are now living in a world where we are all the public by default and must use effort to remain private. For others this is cause for a conversation about what privacy means in the 21st Century. Who can access our information and how much should we be notified when someone is looking for us online.