Resisting the rise of facial recognition

It now seems impossible to stop anyone from privately building up large facial-recognition databases from online photos. But in July, researchers at the University of Chicago in Illinois unveiled a piece of software[6] called Fawkes that adds imperceptible tweaks to images so that they look the same to the human eye, but like a different person to a machine-learning model. If people ‘cloak’ enough of their facial images through Fawkes, they say, efforts such as Clearview’s will learn the wrong features and fail to match new, unaltered images to its models. The researchers hope that photo-sharing or social-media platforms might offer the service to protect users, by applying the software before photos are displayed online.

Calls for regulation

In September 2019, the London-based Ada Lovelace Institute, a charity-funded research institute that scrutinizes AI and society, published a nationally representative survey[7]3 of more than 4,000 British adults’ views on FRT. It found that the majority of people supported facial recognition when they could see a public benefit, such as in criminal investigations, to unlock smartphones or to check passports in airports. But 29% were uncomfortable with the police using the technology, saying that it infringes on privacy and normalizes surveillance, and that they don’t trust the police to use it ethically. There was almost no support for its use in schools, at work or in supermarkets. “The public expects facial-recognition technology in policing to be accompanied by safeguards and linked to a public benefit,” the survey concluded.

Many researchers, and some companies, including Google, Amazon, IBM and Microsoft, have called for bans on facial recognition — at least on police use of the technology — until stricter regulations are brought in. Some point admiringly to the GDPR, which prohibits processing of biometric data without consent — although it also offers many exceptions, such as if data are “manifestly public”, or if the use is “necessary for reasons of substantial public interest”.

1 2 3 4 5 6 7 8 9 10