By their faces shall ye recognize them…
Following on from my post on May 7 about some of the issues raised by the arrival of easy-to-use biometrics in the form of Russian app FindFace, which has identified thousands of people on the highly popular VK social network in Russia.
Since then, events have moved quickly, thanks largely to Russia’s lax privacy laws and the popularity of VK, which has more than 200 million users. As a result, FindFace’s creators, Artem Kukharenko and Alexander Kabakov, aged 26 and 29 respectively, now have more than half a million registered users and more than three million searches, and have been offered a contract with Moscow City Hall to use the app to scan images captured by the 150,000 CCTV cameras dotted throughout the Russian capital. In less than two months, from a purely practical perspective, anonymity is no longer possible in Russia.
The app is able to compare a photograph with a database of one billion images in less than a second, using a personal computer, identify with a 70% success rate, and also find 10 similar faces. This is an app that might appeal to identifying somebody in a snap you’ve taken if they’re on VK so you can contact them, or even to find somebody who looks like somebody famous. Obviously, it will interest the police, who will now be able to locate somebody associated with a closed case, for example. Little wonder it’s being called “a Shazam for people”.
In reality, FaceFind is simply a way of obtaining data to teach a facial recognition algorithm that showcases the technology created by N-Tech.Lab. Use is free for up to 30 searches a month, a policy designed not to make real money, but more to prevent its servers buckling.
One of the founders, Artem Kukharenko, has been working for more than a decade on deep learning, and when he met Kabakov, a philosophy graduate and responsible for marketing FaceFind, they won a competition organized by the University of Washington, MegaFace, in which they scored a 73% success rate in identifying faces from a database of half a million from more than 20,000 users, compared to the 70% scored by Google’s FaceNet.
The key is having the data to teach the algorithm, which has come from VK, given that Facebook was not prepared to cooperate. But the reality is that it is now possible to put together a data base from all kinds of online sources: check it out yourself by Googling your name to see how many photos of you come up — I know that there’s no chance of me escaping identification! Similarly, the authorities will be able to use their vast data bases from security cameras.
FaceFind is now looking to develop specialist uses such as security and marketing: from casinos trying to keep out people they have banned, to all kinds of official organizations, and even companies that want to send advertising to people who have stopped in one of their stores to look at a particular item. At the same time, governments will want to use the service to monitor activists, suspected terrorists, demonstrators, or any activity they want to monitor or block.
Given that this is an app that can be used on any smartphone, we may soon find ourselves not needing identity documents, as police or other officials of the state will simply scan our faces.
Kabakov says that we must now accept we live in a world where our movements are monitored in real time: there is no going back, so all we can really do is to make sure that this kind of technology is as transparent as possible. The debate is already underway in Russia as to the implications of facial recognition, and will doubtless soon spread to other countries as the app is used around the world.
Once again, we find ourselves facing a science fiction future along the lines of Minority Report in which we are observed 24 hours a day, seven days a week in real time. The face is now more than simply the mirror of the soul: it is our permanent and hard to change identity card, for better or worse, or inevitably.
(En español, aquí)
