Perils of face-recognition

The technology has heightened privacy concerns

Three information-technology giants ― IBM, Microsoft, and Amazon ― have decided to stop selling their respective face-recognition (FR) systems to American police departments and federal agencies. This decision was driven by opposition from employee-shareholders. The deployment of FR at protests against the killing of George Floyd may have hastened these self-imposed moratoriums, but it may trigger retaliatory action. US President Donald Trump has tweeted in favour of a ban on awarding government contracts to Microsoft. Amazon was already on his hate list, because Amazon Chief Executive Officer Jeff Bezos is the owner of The Washington Post newspaper and Amazon has a one-year moratorium on selling FR in place. FR systems have been the focus of privacy debates for years. While it is cited as a convenience for card-less financial transactions, or international travellers who can be waved through immigration, it can be used for more sinister purposes.

Many tech companies have faced pushback from within their own ranks about deploying morally ambiguous technology including FR and other AI-driven R&D that could be deployed in futuristic weaponry. Opposition in Silicon Valley hardened circa 2018, when it became apparent that FR could be deployed for facilitating targeted assassinations by drones. Amazon and Microsoft have both pledged to avoid developing military FR. Mass surveillance is another obvious application for FR. Police in Hong Kong, the UK, India, and in the recent Black Lives Matter protests have been routinely deploying it to identify activists. Mainland China also uses FR to enforce its social credit system, denying citizens with poor social credit scores the right to board flights or trains. It is believed to have also used FR to assist in contact tracing during the pandemic. Activists in tech-savvy Hong Kong have developed various methods of blocking FR during the long agitation. This includes low-tech ploys like wearing T-shirts featuring faces, motorcycle helmets, shiny jewellery, and odd streaks of face-paint to confuse FR. The pandemic has also had technological consequences. On the one hand, people are routinely wearing masks. On the other hand, cutting-edge FR can now identify people with a degree of confidence even with the lower half of their faces covered.

It is not clear if these corporations will continue to research FR and supply their systems to agencies outside America. Other, less-ethical companies could also move into this space. Given a proliferation of CCTV surveillance, any individual going about a normal daily routine may be recorded multiple times on any given day. If that footage is run through FR systems, privacy is severely compromised. In several US states, including California and Illinois, FR usage by law-enforcement is banned by legislation. In the EU, the use of FR falls under the General Data Protection Regulation. Citizens can ask for footage held by either a private organisation, or the government, to be destroyed under the “Right to Forget”, unless acquired with consent and used for a specific purpose. No such moderating legislation exists in India, where the government has a blanket right to conduct surveillance, even under the pending proposed Personal Data Protection Bill. FR systems have been deployed at many of the anti-Citizenship Amendment Act protests. This is one of many areas of concern for civil rights activists in India.