BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Hello, Stranger

This article is more than 10 years old.

Science fiction writers have long imagined a future in which facial recognition technology makes anonymity in public obsolete. A research study at Carnegie Mellon University suggests that this Minority Report future has already arrived, thanks to facial recognition products now commercially available, combined with the 750-million-person identity database called Facebook.

A CMU research team led by associate professor Alessandro Acquisti took candid photos of 93 random students on campus using a $35 webcam. Within seconds the researchers were able to determine the identities of a third of their photogenic guinea pigs, using off-the-shelf facial recognition software from PittPatt, a software company recently acquired by Google, and publicly available profile photos from Facebook. The researchers had an even higher rate of success using the same technology to identify more than 100,000 Pittsburgh singles with otherwise pseudonymous accounts on a dating site, adding yet more complexity to the world of online dating.

"We've been thinking about privacy in cyberspace, but we need to start thinking about it in the real world, with augmented reality," says Acquisti. He was also able to determine the first five digits of some of the identified students' Social Security numbers based on previous research he has done showing that those numbers are predictable if you know a person's birth date and birthplace. That information is usually available on a person's Facebook profile page.

It seems that Aldous Huxley was right and George Orwell was wrong. Ubiquitous surveillance isn't being orchestrated by the governmental Big Brother of 1984 but by advances in technology designed for convenience's sake adopted eagerly by private citizens. Acquisti calls it the "democratization of surveillance." And it's coming fast. Soon after the riots broke out in London at the beginning of August, a technophile band of vigilantes formed a Google Group to discuss applying the methods pioneered by Acquisti's research team to online photos and videos of rioters, in order to help identify looters for prosecution by law enforcement.

Facial recognition tools identify a person by analyzing dozens of features, such as the length of a forehead and the distance between the eyes and the nose. Google, Facebook and Apple have already made them freely available for people to tag their friends and family in photo albums. But at what point will you have the option to snap a photo of a stranger and then pull up his or her name and whatever information the Internet offers up about them?

The capability is there, but the willingness to unleash it isn't. Yet. Google has said that its image search tool, Goggles--which uses a photo instead of text as a search term, letting you snap a photo of a painting, for example, and find out who its creator is--has the capability to find people's identities based on a photo. But the search giant had chosen not to enable the feature because it was deemed too creepy.

When Facebook recently rolled out an internal facial recognition tool allowing friends to automatically tag other friends in photos, it inflamed some privacy advocates (whose combustion levels tend to be high around anything new that Facebook does). Those advocates seemed to overlook the fact that the very existence of Facebook's extensive database of photos linked to people's names creates the possibility for third parties to build a facial biometric print. While Facebook's terms of service prohibit the scraping of information from its site, there are no technological tools in place to prevent it. And, even if there were, scraping is certainly not limited to Facebook. "For most of us there are already identified photos of us somewhere on the Internet, and this trend will only increase," says Acquisti.

The CMU researchers wrote a facial recognition iPhone app but decided not to release it to the public. "We developed the app to make a point," says Acquisti. "A future in which we infer sensitive information from faces is already here, and we better start thinking whether we really want to live in that future."

Their tool, though, would be valuable only for those seeking to identify the 25,000 CMU students whose photos the researchers grabbed from Facebook. We're still a few years away from an app that would work on a national or global scale. The two main hurdles, Acquisti says, are finding a large and reliable source of facial images (hello, DMV!) and securing enough computing power. Comparing a photo to a database of, say, 300 million photos would take many hours, whereas a functional app would need to do it in ten seconds. "If I had to guess, I would say that we are less than five years away," says Acquisti.


Specialized facial recognition products are already on the market. BI2 Technologies has developed a $3,000 iPhone add-on called MORIS (Mobile Offender Recognition and Identification System) that dozens of police departments across the country started adopting this year. Police officers can use MORIS to take photos of suspects, and the app, using iris scans and facial recognition, will identify anyone who already has a photo in a criminal database.

Microsoft and Intel have worked together to create digital billboards with face-detecting cameras that pick up on the gender and age of passersby to show them relevant ads. A startup named SceneTap installed similar cameras in bars in Chicago earlier this year, so that Windy City nightlife lovers can use their smartphones to check the crowdedness, gender ratio and average age of patrons at a given bar before deciding where to go.

So far law enforcement officials are tentative about using the technology to identify people not already in criminal databases. Though a Canadian insurance corporation graciously volunteered the use of its driver's license photo database to Vancouver police after the devastating post-Stanley Cup hockey riot there in July, authorities declined the offer after privacy advocates expressed concerns about violation of Canadians' civil liberties. In London, Scotland Yard is using facial recognition tools along with tips from the community to identify rioters.

As good as the technology has become, it still has accuracy problems that can cause significant harm. A Massachusetts driver sued his state DMV after his license was wrongfully revoked because he looked too much like another driver, leading to ten days of administrative wrangling to get it reinstated. "We are not the beautiful and unique snowflakes we think we are. There are many people who look like us," says Acquisti. More than 30 state DMVs already apply facial recognition algorithms to driver's license photos to prevent identity fraud.

If you're worried about your privacy, now might be a good time to change your Facebook profile photo to one in which you're wearing ridiculously oversize sunglasses.

Special Offer: Free Trial Issue of Forbes