A facial recognition app has been used to maliciously unveil the identities of Russian sex workers. Digital rights experts have warned that such technologies may undermine any semblance of privacy in the near future.
As if torn from the pages of dystopian fiction, an unwitting Russian company in February launched a facial recognition app that linked to the country's largest social network VK, formally known as VKontakte, effectively planting the seed for a gross intrusion of privacy.
The facial recognition platform FindFace gained notoriety at the beginning of April when St. Petersburg-based photographer Egor Tsvetkov captured images of unsuspecting commuters on public transport, and used it to discover their VKontakte profiles.
"The ability to quickly search for an anonymous person's page in the social network, in seconds converts an impersonal object to a close friend," Tsvetkov said of his project "Your Face Is Big Data."
"My project is a clear illustration of the future that awaits us if we continue to disclose as much about ourselves on the Internet as we do now," the photographer added
Tsvetkov posted images of people he photographed taking the metro alongside one's from their VKontakte profile
Sex workers targeted
However, while Tsvetkov's project aimed to prove a point about the application of open source technologies, users of the Russian image-board Dvach had other intentions.
By mid-April, a group on Dvach took to the app FindFace to correlate images of Russian sex workers and porn actresses to their personal social media accounts, reported security software company Sophos.
The women's personal photos, profile details and contact information obtained from their profiles linked their personal identities with their professional lives when they were consequently posted online.
FindFace founder Maxim Pellin told Russia's TJournal that the application was not meant to undermine privacy, but instead to make it easier to discover new friends on social media networks.
"We make every effort to protect all network users from potential malicious actions and are ready - in case of need - to give all the information to search for such users," Pellin said.
Meanwhile, Artem Kuharenko, CEO and co-founder of NTechLab, the company that developed the software, told DW that updating your privacy settings would effectively opt you out of the software’s algorithm.
"FindFace only goes through the open profiles on VK. If you want to be excluded from the identification, the only thing you need to do is just to make your account private in the settings," Kuharenko said.
However, Christian Funk, head of global research at software security group Kaspersky Lab, told DW one of the issues posed by FindFace's integration with VKontakte is that the restrictions on privacy are limited.
"In that social network, it is not possible to restrict access to your avatar or profile picture," Funk said.
While others may have used FindFace for other reasons, Kuharenko said that the purpose of the software implemented in FindFace was to facilitate connections between people based on their looks.
"Our friends came to us with an idea to use our facial recognition algorithm to create a new type of dating service, where people would be able to find partners by their appearance," Kurahenko told DW.
"That was a great way to demonstrate the abilities of our platform, and that is how FindFace appeared. We work closely with them on further FindFace development," he added.
Off to other platforms
But FindFace isn't the only application of facial recognition software. In December, computer science professors from the Seattle-based University of Washington announced the use of a data set obtained through Creative Commons-licensed photographs on Yahoo's photo community and social network Flickr.
The reason for using Flickr was to "enable benchmarking and distribution within the research community," according to a conference paper written by several computer scientists, which included Ira Kemelmacher-Shlizerman and Steven Seitz of the University of Washington's Department of Computer Science and Engineering.
"While recent face data sets have leveraged celebrity photos crawled from the web, such data sets have been limited to a few thousand unique individuals; it is challenging to find a million or more unique celebrities," the paper said.
"Instead, we leverage Yahoo's recently released database of Flickr photos. The Yahoo dataset includes 100 million creative commons photographs and hence can be released for research," the paper noted.
Leveraging consumer technology
As consumer imaging technology becomes increasingly ubiquitous and seamlessly integrated into everyday life, their growing quality when paired with other technologies, such as facial recognition software, has incidental consequences.
For example, the expanding image resolution and dynamic range in smart phone cameras provide a caliber that was previously only found in DSLR or mirrorless cameras.
"The quality of the cameras in modern smart phones is good enough so that the facial recognition software actually does a decent job" at returning positive matches, Kaspersky Lab's Funk told DW.
Funk's comments come as technology giant Sony earlier this week submitted a patent application for a contact lens that carries a camera fitted with aperture control, in-body image stabilization and a storage drive in an object that would be no larger than a portion of your eye.
'Cumulative effects': An end to privacy?
The technological convergences of consumer technology and open source facial recognition software pose major issues for privacy rights, not only in Russia, but also in Europe and other parts of the globe.
European Digital Rights Initiative (EDRi) Director Joe McNamee told DW that facial recognition software is only one part of the puzzle, adding that it becomes more so "intrusive" when merged with information obtained from other sources.
McNamee said these include social media data, online tracking data and photographs taken "without either active or passive consent."
"A lot of the data generated from adding data sets to data sets are guesses that don't need to be correct, they just need to be correct enough for the business purpose for which they are generated," McNamee said.
"In a world of 'big data,' it is a mistake to consider individual technologies in isolation. We really need to start thinking about the cumulative effects," McNamee stressed.