Subscribe to read:

Facial recognition’s risks demand a temporary halt

Upgrade your account to read:

Facial recognition’s risks demand a temporary halt

Artificial intelligence

Facial recognition’s risks demand a temporary halt

The EU should define where controversial technology can be deployed

A ban on facial recognition is only one part of a broader review that so-called surveillance capitalism badly needs © Bridget Bennett/Bloomberg

Brussels and Silicon Valley rarely see eye-to-eye when it comes to technology regulation. But on facial recognition, EU regulators and Sundar Pichai, chief executive of Google parent Alphabet, appear to be moving in the same direction. Mr Pichai this week backed a temporary moratorium on the technology — an option under consideration by EU regulators as part of a broader strategy on artificial intelligence. Brussels should follow through on this idea. A new technology with a considerable risk of unintended harms should not be rushed out.

A potential temporary EU ban on facial recognition in public spaces was described in a draft white paper on AI seen by the Financial Times. During this hiatus, regulators would consider how to assess the impact of the technology and mitigate its risks. There is a precedent for this in the US, where the cities of San Francisco and Oakland have banned the use of facial recognition by public bodies.

While the draft proposal mentions the concern that a delay would hamper development, these are outweighed by the risks of a reckless rollout. In the western Chinese province of Xinjiang, security services have used facial recognition to track the Muslim majority population. Even in democratic countries, facial recognition raises serious questions about the public’s ability to opt out of surveillance which is built into urban environments. An investigation into Clearview AI, which scraped photos from millions of websites to create a database for police, offers an unsettling vision of the state of modern privacy.

The accuracy of facial recognition has also been called into question, particularly with regards to the performance of systems when identifying women and non-white faces. Yet around the world, government and law enforcement agencies have poured money into acquiring the technology. Given the risk of misidentification and false positives, there is a cause for alarm.

Emotional recognition, an offshoot of the technology which claims to be able to identify an individuals’ mood, is even more dubious. Yet it is already being used by recruiters and border agencies on the assumption that it can discover bored applicants or nervous passengers.

A temporary moratorium would not simply allow developers to improve the technology and minimise public exposure to incomplete and faulty systems. It would allow the EU to decide which settings are appropriate for using facial recognition. Debates around AI and ethics have produced dozens of frameworks around the world, whose suggestions are often anodyne rehashes of “do no evil”. Banning the system from certain surroundings would offer far more practical guidance. While facial recognition may be warranted on the grounds of national security, for example, it is difficult to justify its use by enterprises such as retailers. Private uses are often more obscure, and should require a higher bar for consideration.

A time-limited European ban on public uses of facial recognition will not end its development in the US and China. It is also only one part of a broader review that so-called surveillance capitalism badly needs. Google, which plans to gather vast amounts of healthcare data from millions of users through sensors in wearables, mattresses and even toilet seats, is as implicated in this as any of its rivals.

Nevertheless, Europe has a clear opportunity to show other jurisdictions that the importance of protecting citizens should constrain the tech industry’s rush to create the next great Big Data product.

Letter in response to this editorial comment:

Companies can adopt a voluntary moratorium on facial recognition / From Carly Kind, London, UK

Copyright The Financial Times Limited . All rights reserved. Please don't copy articles from FT.com and redistribute by email or post to the web.

Content not loading? Subscribers can also read Facial recognition’s risks demand a temporary halt on ft.com