Tech giants are building vast facial recognition databases – and all of us should be concerned about the implications
Worried about your selfie being added to an internet database? You should be
The unlikely viral hit of last week was Google’s Arts & Culture, an app that lets users explore the world’s most fascinating museum collections from the comfort of whichever chair they happen to be sitting in. Needless to say, this wasn’t down to a surge of global interest in the arts.
What suddenly drew people’s attention was a recently implemented feature (in certain territories) called “Is Your Portrait In A Museum?”, where Google attempts to match your selfie snap with a painting stored in its database.
Over 30 million people rushed to download the app in order to observe their uncanny likeness to Frans Hals’s Laughing Cavalier or if they were having a particularly bad day, the poor soul depicted in Edvard Munch’s The Scream.
It’s the kind of innovative service that sends the internet wild every so often, but enthusiasm was tempered by voices warning of a possible ulterior motive by Google. Actress Alyssa Milano summed up the feelings of many people in a tweet: “Anyone suspicious of just surrendering your facial recognition to Google,” she wrote, “or are we confident they already have that at this point?”
Her point, slightly clumsily made, is that pictures of our faces have acquired a value in recent years, and encouraging us to take photographs of ourselves and upload them to the internet isn’t necessarily an act of benign generosity. Snow, a Korean selfie app that Facebook once showed an interest in buying, received $50 million (Dh183.65m) in new investment this week; the company stated that this will be partly used to develop facial recognition technology.
The more selfies you throw at a facial recognition system, the better it gets. For its part, Google emphasised in a blog post that it doesn’t use your selfie for anything other than portrait matching, and “only keeps it for the time it takes to search for matches”. But even if we take Google at its word, the broader concern amongst privacy advocates is that we’re sleepwalking into a situation where our happily-surrendered pictures end up being used for other purposes, the most obvious of which is to identify us. Of course, the practise of giving pictures of our faces to technology giants such as Apple, Facebook and Google (and then telling them who’s in those pictures) has already become thoroughly normalised, but this is not a trend restricted to smartphones.
A new report by researchers at Georgetown University criticises the practice of facial recognition scanning by airlines at American airports, describing it as “a serious escalation of biometric scanning of Americans” with “no codified rules that constrain it”. In the UK, a pilot scheme was recently launched to install facial recognition software at the checkouts of two national supermarket chains, in order to instantly approve the purchase of age-restricted goods.
In Japan, sources close to the Tokyo 2020 Olympics have indicated that facial recognition may be used to identify athletes, officials and journalists attending the games. Each of these data collection points markedly improves the accuracy of the recognition service that powers them.
There’s no question that facial recognition systems can bring added convenience – although the fact that they’re not 100 per cent accurate puts question marks against their use for security purposes. But one of the major concerns surrounds the question of who, one day, might be given access to that data. The Georgetown University researchers estimate that half of all Americans already have their features captured in a facial recognition database, but those people have no idea which one, how they might have got there or what they might eventually be used for.
The iPhone X, discussed in these pages a few weeks ago, continues to be of interest to privacy campaigners in this regard; they continue to seek assurances over how much facial data accumulated by the phone is revealed to third-party app developers.
Aside from the organisations granted access to databases of our faces, there are those who simply hack their way in.
Woodrow Hartzog, Professor of Law and Computer Science at Northeastern University, voiced his concerns to the website Gizmodo: “If a database of faceprints were compromised,” he said, “it would have a ripple effect on authentication systems that used the faceprint, as well as possibly allow unauthorised parties to make use of the faceprint for surveillance.”
In India it was recently reported that access to personal information of more than a billion citizens – including their photographs – was being offered online for less than £6 (Dh31), leading to questions over the safeguards that surround identification data held not just by private companies, but by governments, too.
The potential ability of authorities to instantly identify citizens from a single image is the sci-fi scenario that worries privacy campaigners most of all. Last September, Moscow unveiled a new system linking the city’s CCTV cameras with a facial recognition system designed to locate and track criminals. While limits on computer power currently make it impossible for every citizen to be tracked, overcoming those limits would simply be a matter of time.
In China, the government’s ambitions have been more openly stated; their “Xue Liang” plan (translating as “Sharp Eyes”) seeks to connect CCTV networks and databases to create a national surveillance platform. Chinese authorities have already come in for widespread criticism following the introduction of facial recognition systems in the Muslim-dominated Xinjiang region. The system alerts authorities when certain citizens stray outside particular areas.The lack of effective regulations and laws across the globe against the accumulation of facial data has meant that databases have grown unchecked. Canadian deep-learning guru Yoshua Bengio publicly warned about the ethical implications of this drift back in October, but pushbacks are rare. One company, D-ID, did announce earlier this week that it has received $4m in investment to assist with development of its tools that block facial recognition algorithms from analysing pictures and videos – but with selfie culture and CCTV systems generating billions of pictures every week, it has a gargantuan task on its hands.
The best that we can do, as users and citizens, is to remember that selfies might seem throwaway, but they do have a value – even if it does take Alyssa Milano to remind us of the fact.