FastCompany Reveals How Anonymous Data Location Identifies Us

FastCompany asked a data scientist to see how much he could learn about two individuals using their “anonymized” geo data from Google Maps.

The short answer? A lot.

With more digging online, Lotan filled out the rest of his profile, which included an enthusiasm for hiking, cafes, and the Vista Ridge Community Center in Erie. “On Facebook, you can see everything else, like their kids and a puppy.”

By now, Lotan wasn’t just keeping track of the places this person frequently visited using anonymous smartphone location data: He had managed to crack their entire identity.

If a malicious actor were to obtain this GPS data–collected by any number of smartphone apps, and collected by big companies and startups, advertisers, and law enforcement, with little oversight–they could use it to manipulate or harass that person, or worse.

And then there’s this:

In 2013, researchers at MIT and the Université Catholique de Louvain in Belgium published a paper reporting on 15 months of study of human mobility data for over 1.5 million individuals. What they found is that only four spatio-temporal points are required to “uniquely identify 95% of the individuals.” The researchers concluded that there was very little privacy even in raw location data. Four years later, their calls for policies rectifying concerns about location tracking have fallen largely on deaf ears.

Clearly, there’s some legislation work that needs to be done about this. But more importantly, the article lists a few easy ways you can protect yourself from being found online, whether you’re using an iPhone or Android device.

I also don’t want to be the boy who shouted wolf — at least, not yet. Much of this is about protecting your data from advertisers, much of whom have obvious intent. But the real problem here is protecting yourself from hackers and other malicious actors who would be interested in obtaining your geo-location for more nefarious reasons.

Nobody is saying that these location-based features aren’t useful (at least, I’m not). But there’s a lot to be said about being careful about who you trust with this data.