Your face may be your most significant privacy concern


110 days ago

Machine learning related to facial recognition may overtake the risk to privacy that is currently centred on cookies that track your browsing.

If the sentence above does not make sense, let me take a step back.

A cookie is a piece of code used on web browsers to perform particular functions. Most are useful and necessary like saving you from having to log in to a password-protected site each time you click a link.

Some are used to track your behaviour online and that too can be useful if you would like to have a site remember personal settings without requiring you to create an account.

But it also allows companies that use cookies to learn more about you. As you are reading this a cookie set by Google allows us to know which pages you have read on this site and when last you visited. This allows us to know which articles are popular and how often you might read our content. We don’t know who you are personally, but we can see than some subjects get more attention than others and so we can look to make more of that sort of content or less of it if very few bother reading it.

Companies like Google and Facebook also use other kinds of cookies that are stored on your computer than are accessed when you visit other sites which allow Google, Facebook and others to know which sites and stories you have read. If you search for a product and then see ads on other sites which have ads of the product you were just searching for is how they manage to do that.

An ad that is of a product you might actually be looking for is helpful, but the concern is when enough data about your habits is known, you may be more susceptible to manipulation.

Perhaps you were trying to restructure your debt following the lock-down and did some searches for potential options after you are done, you may choose to play a relaxing online game. In that game which is probably free to play are ads and those ads display specific ads related to what Google knows about you and what kind of person an advertiser wants to reach. Usually, there is an auction with the advertiser that bids the most getting to display their ad. A get rich quick scam may wish to bid quite a bit for those that need to get out of debt in a hurry and by knowing that might get you to click a link while playing an innocuous game that could lead to you using the last of your money when you would have been less likely to have chosen if it was listed with other search results.

The good news is that companies like Apple don’t want their device owners to get caught out and so limit how long a cookie can remain active and don’t allow cookies to be set by others. Google is following suit and many reputable advertisers also want to see it better regulated in order to avoid more legal regulations.

Unfortunately despite the improvements, we will need more transparency and limitations placed on ad companies via a country’s laws rather than simply trust that the industry will look after itself.

But even as that chapter begins to draw to a close thanks to improvements, a new method (if we don’t get the regulations in place at the same time) might be far more powerful than trying to work what we like from our online activity.

It is our faces.

We take a lot of pictures of ourselves. Children grow up in front of cameras and the images live on the internet. Humans are quite good at making quite accurate assumptions about people from just looking at them. We can determine gender and age pretty well, we can guess things like personality with greater than 50% odds, but for all our skill, computers trained to look for the differences between introverts and extroverts can do an even better job.

Dr Michal Kosinski is an Associate Professor of Organizational Behavior at Stanford and has been researching the influence of personality on performance. That seems fine, but his research suggests that we can determine personality from facial features and if personality can predict performance it may lead to a job interview consisting of just your picture.

It may be early days, but typically we wait until too late to try to restore something that is lost. The one asset that perhaps we should try to keep the most private is our face. Once marketers know what we look like, they don’t really need much else to begin making very educated guesses about how to sell us stuff. Politicians could determine how best to influence us and governments might decide to mark us out for surveillance based on how we look.

We might be promised that it could help find serial killers just by looking at them and it may, but it will also be used for lots of other things we are not informed about and so we need two things before it goes too far.

We need transparency whenever it is being used. If you are shown an ad it should say that the ad has been selected based on what is known about you. Facebook knew that it could influence users with what they decided to show in the news feeds. They actively allowed advertisers to target groups based on their likes and personality attributes, but after the Cambridge Analytica scandal they have made all ads public and the ad will reflect what targeting was used to show you the ad.

The second may sound crazy, but we need a non-real face. We have used names other than our own for the longest time, but it will not be our name that gives us away as much as our face.

The website, this person does not exist generates a wide range of realistic-looking faces constructed by the same machine learning computers that can determine who we are. Using a generated image of ourselves will buy us some time, just be careful to not choose a face you like too much, the machines may be able to determine as much about you from the kind of face you choose as from your own.

This article first appeared on 702 : Your face may be your most significant privacy concern

Recommended

by NEWSROOM AI

Recently heard on air

Nea - Some Say (Felix Jaehn Remix)

LISTEN TO 947 LIVE
ON AIR Essential 947 NOW PLAYING Nea - Some Say (Felix Jaehn Remix)
LISTENING TO
Recently played