The advent of computer vision and facial recognition technology can be largely credited to the advances in machine learning, more broadly — artificial intelligence (AI), and its associated technologies. Training datasets that feed machine learning algorithms have gotten larger, processing power has improved and become cheaper via cloud computing, social media usage is at an all-time-high and camera resolution continues to be enhanced.
During the ongoing protests in Hong Kong, protesters were seen to be using laser pens to disrupt and damage facial recognition cameras at the Chinese government liaison office.
This feature of the pro-democracy demonstrations drew China’s culture of surveillance back into sharp focus. It is common knowledge that China profiles and surveils Uighur muslims in the Xinjiang region and uses big data and AI under the guises of necessary “security” and “stability”, without unified privacy or data protection laws. An estimated 170 million CCTV cameras are operational across the country with an additional 400 million due to be installed in the next three years. Initial testing of glasses utilising built-in facial recognition technology has also been conducted in populous cities like Zhengzhou, allowing wanted criminals to be identified by the AI in ~100 milliseconds.
That said, China isn’t the only cause for concern and while it is easy for national governments to condemn this sort of behaviour on the world stage, the UK is in the trial phases of using facial recognition technology (and doesn’t have policies in place to govern its usage) and in the USA, roughly half of American adults are in a law enforcement face recognition network according to a Georgetown Law report from 2016.
Amazon’s facial recognition offering — “Rekognition” — has been shopped around to government agencies in the USA, but has been held up by shareholder misgivings and a challenge from the American Civil Liberties Union (ACLU). Similarly, Microsoft’s Face API has been hamstrung by a lack of federal oversight and legislation to ensure its proper use, leading to their president suggesting a number of safeguards. Facebook has used some form of facial recognition in its photo tagging for some years and it seems they are using the massive datasets to power an AI that creates “templates” of users, which could conceivably lead to them landing in more hot water regarding privacy. Google have adopted a cautious attitude before they decide to commercialise their facial recognition tech, citing the need for “responsible development”.
UPDATE: Facebook faces a class action lawsuit over breach of Illinois’ Biometric Information Privacy Act which could lead to $35 billion payout.