Citizens across the world are increasingly losing their right to decide who and where their images are used, thanks to data mining and facial recognition technology. In the US, a little known AI company, Clearview AI, has amassed three billion images from people who know nothing about it, writes Kashmir Hill in her chilling book, Your Face Belongs To Us. Supplied
ANONYMITY no longer exists.
A US company has secretly amassed 30 billion images of people across the world without their knowledge and upload 75 million new photographs to its database daily.
Clearview AI has developed a powerful app capable of identifying almost anyone whose photo and name has appeared on the internet, even when they are wearing a hat, glasses or a mask.
Journalist Kashmir Hill, who details the chilling information in her book, Your Face Belongs To Us, says Clearview AI has more images than the amount of people on Earth and, for many individuals, the company has different versions of their face.
By simply uploading a photo, the database will show you all the places on the internet where that person has appeared. It can also reveal your name, where you live, who your friends are, your social media profiles, and even photos on the internet that you didn’t know about.
What’s more, Hill says, they are developing augmented reality glasses that will identify anyone to the user just by looking at them.
Speaking to the Independent on Saturday this week, Hill said: “All of this information that has been collected about us over the last few decades can now be attached to our face in the real world. I think that is very chilling. The face is so personal and just this idea of being able to not be anonymous anymore in our kind of movements in the public world is very troubling.”
Hill, who found more than 100 pictures of herself on the database, says everyone is in there, even people in South Africa, raising a multitude of ethical, privacy and safety concerns.
For instance, an undercover police officer could be identified, blowing their cover, and people in witness protection were also at risk.
“We’ve already seen the CIA issued a warning to their operatives around the globe that they need to be more careful with their confidential informants because they are being identified using artificial intelligence including facial recognition.”
She says that as a journalist, she has to take extra care to protect her sources who could be identified through the app.
The technology is used only by law enforcement agencies in the US, especially in child abuse cases, but Interpol has also expressed an interest.
Australian Hoan Ton-That,who started out developing Facebook quiz apps like “Have You Ever” and “Would You Rather” to fund his stay in the US, is one of the masterminds behind the company.
Hill said initially Clearview AI was “respecting” international privacy laws which enabled people to access and delete information about themselves but the company recently told her it would no longer allow that.
“And so, unless you live in one of the few states in the United States that does give you the right to access and delete the information a company holds on you, there’s nothing you can do: you can’t get out of Clearview’s database,” she said.
They have images of her in the background of someone else’s photo, and photos taken two decades ago that she didn’t know were on the internet. Even when she covered her nose and mouth with her hands, it was still able to identify her.
But she warned that the technology could make mistakes, especially with people who look similar to each other.
In a case of mistaken identity, one man spent a week in jail and was then released because he had a mole on his face that the suspect didn’t.
“This technology can identify doppelgängers, it can lead to kind of mistaken harassment, mistaken arrest, if it is not kind of treated with a certain kind of scepticism,” said Hill.
While it have the biggest database in the world, there were copycat companies following the Clearview model of scraping the internet, collecting photos and making them searchable by face, really reorganising the internet of photos around people's faces, said Hill.
Her investigations revealed that the company’s aim was to sell the database to anyone willing to pay for it; it was a fluke that the police were the only ones with access to this.
Companies with similar technology offer a subscription model of about $30 (R560) a month which entitled you to 25 searches a day, but without any technical measures in place to ensure you can only search for your own face.
Hill said they were essentially competing with Clearview to “strip all of us of our anonymity”.
She said the companies believed they had the right to use any information found on the internet and despite push-back from privacy regulators in Canada, Europe, Australia and other parts of the world, enforcement has not been effective.
In her book, Hill writes that Ton-That attributes his journey from building banal Facebook apps to creating world-changing software by standing on the shoulders of giants because he would not have been able to build the technology from scratch.
“Those giants, computer scientists who had toiled in academic labs and Silicon Valley offices, had paved the way not just for Clearview but for future data-mining companies that may come for our voices, our expressions, our DNA, and our thoughts.
“They yearned to make computers ever more powerful, without reckoning with the full scope of the consequences. Now we have to live with the results,” she writes.
Hill is a tech reporter at The New York Times where she writes about the unexpected and sometimes ominous ways in which technology is changing our lives.
Your Face Belongs To Us is a paperback book which retails for R410.
The Independent on Saturday