This is Scientific American 60 Second Science.

I'm Larry Greenmin.

When you install an app on your smartphone,

you're often asked whether you'd like to share your list of contacts with that app.

That might be a convenient way to connect with friends and family likewise using

say Instagram or what's app, but it also means you're giving away their

personal information to the app developers.

And that personal info could end up being used to create so-called shadow profiles of your contacts

even if they don't use that app or social media service. Shadow profiles emerged as a

potential problem in 2011 when an Ireland-based advocacy group accused

Facebook of gathering information on non-users,

including names, email addresses, phone numbers, and physical addresses.

The following year, researchers showed that social network companies such as

Facebook could use machine learning to pretty

accurately predict whether two non-members known by the same member also know one

another. Not exactly Big Brother, but a recent study in the General

Science Advances raises the stakes.

In that work, David Garcia, chair of systems design at the Saitek University E.T.H. Zurich

used the social network members personal information to infer relationship status and sexual orientation of the members contacts who did not have their own user accounts on that social networking site. He was able to do that using, of all things, data from the now

defunct Friendster social networking site. He says he chose those two attributes, relationship status

and sexual orientation, because they can carry important privacy consequences

and were both available in the Friendster data set.

...