Civil liberties groups are warning police use of facial recognition technology (FRT) is “out of control,” but despite well-documented problems with accuracy and discrimination, regulators appear reluctant to rein it in.
Discuss or suggest changes to this story
Discuss or suggest changes to this storyTalk
In the United States, federal law enforcement agencies have access to pictures, via driving license records and ID photos of around 117 million adults, according to the Center for Privacy and Technology, based at Georgetown University.
In a study released in 2016, the center found that enforcement agencies have access to these records based on their ability to use FRT to assist with policing. The same study showed FRT was being employed with little regulation or oversight.
This report “comes the closest to giving an overall understanding of how widespread this technology is,” said Clare Garvie, one of it authors, “but it’s constrained in that we surveyed a little over a hundred law enforcement agencies, and there are over 18,000 in the United States.
“The deployment is probably more widespread and probably more advanced than we’re even aware of.”
You can edit or expand this story
You can edit or expand this storyEdit
Growing part of everyday life
People are increasingly interacting with FRT in everyday life. Social media feeds identify people to tag in pictures; smartphones group photos by subject.
The benefits of using this technology for law enforcement are immense. A huge amount of personnel hours can be saved if, instead of having people trawl through CCTV footage looking for a known suspect, the video can be automatically checked for a suspect’s face.
Practical issues with this have been well-documented. The first major use of such technology by police in the UK led to a 92 percent error rate. FRT relies on software that “learns” to recognize faces using a database of photos. These databases have often been overwhelmingly comprised of caucasian faces, making the FRT less reliable for identifying minority groups already subject to disproportionate levels of scrutiny from law enforcement.
Many developers say these issues are being addressed and that the technology has become more accurate. Following a software update released on June 26, Microsoft said it had greatly expanded and diversified the database on which its FRT is based. The company said it reduced error rates for men and women with darker skin by up to 20 times.
However, in an article for TechCrunch on June 25, Brian Brackeen, CEO of FRT developer Kairos, said the technology is “not ready” to be used by police.
Brackeen went further, outlining how FRT has been used by authoritarian governments such as China, and arguing that government use constitutes “an extraordinary invasion of the privacy of all citizens,” as it relies on their having access to so much individual data already.
Moving in a regulatory vacuum
Despite these widely acknowledged issues, developers and customers see massive space for business growth for this technology. Axon, one of the biggest suppliers of police body-cameras and Tasers in the United States and UK, announced in April it’s exploring how to use AI tools including FRT in its body-cameras to “increase police efficiency and efficacy.”
“That’s an incredibly dangerous combination,” said Garvie. ”What if this technology makes a mistake? All of a sudden we have as the final arbiter a flawed technology [and] an officer who has a moment to decide whether or not to draw his weapon and act with potentially lethal force.”
In the United States and the UK, critics argue the root of their concern is a shortage of appropriate government oversight.
“The legislature hasn’t caught up,” said Garvie. “It’s being deployed faster than the legislatures are passing laws to address it.”
Megan Goulding, a lawyer at UK human rights group Liberty, told WikiTribune that FRT is effectively being used “lawlessly” in the UK.
“That seems to be a conscious decision by the government because they’ve also made statements that decisions to deploy the technology are operational choices for police forces,” said Goulding.
Freedom of Information requests reveal that several UK police forces have been using FRT, including South Wales, Leicestershire and Humberside. London’s Metropolitan Police began an expanded trial of the technology on June 29, while on June 28 the government announced it would begin using FRT for border controls.
Something missing from the story? Say so
Something missing from the story? Say soTalk
Calls for oversight gain momentum
On June 28, the UK Home Office delivered a “biometric strategy” paper, addressing police use of FRT.
The paper took four years to prepare and did not address the concerns of a parliamentary committee, which complained that IT systems used by police forces are not technically capable of deleting images of innocent people. This means their use of FRT does not meet standards regarding fingerprinting and DNA records.
In the United States, a 206 report from the Center for Privacy and Technology outlined a recommended regulatory framework that would address concerns of most civil liberties advocates. This has been used for consultation in Maryland, according to Garvie, but not yet adopted by federal lawmakers.
With progress stagnating in terms of advancing regulations, campaigners are looking for more direct action.
Liberty is supporting a legal challenge to the use of FRT by police by arguing it breaches the European Convention on Human Rights protection of the rights to privacy, free assembly and expression, as well as protections against discrimination.
Those concerns echo the fears of the American Civil Liberties Union (ACLU), which has been campaigning since early 2016 against the systematic use of facial recognition software by law enforcement.
Recently, this campaign has focused on Amazon’s sale of its Rekognition FRT. The ACLU says the online retailer has been heavily marketing to government agencies.
On June 18, the ACLU delivered a letter to Amazon CEO Jeff Bezos, supported by a petition with 150,000 signatures, arguing the company’s software is “primed for abuse in the hands of governments.”
Following this, Amazon employees wrote to Bezos on June 21 calling on him to stop selling the technology. Amazon shareholders had already sent Bezos a letter, on June 15, citing concerns over the potential use of the software and and knock-on effects for the company’s share price.
Amazon issued a response to the ACLU’s earlier concerns on June 1, stating there had been no reported abuse of the technology by law enforcement.
“[We] believe it is the wrong approach to impose a ban on promising new technologies because they might be used by bad actors for nefarious purposes in the future,” Amazon argued. “The world would be a very different place if we had restricted people from buying computers because it was possible to use that computer to do harm.”
While some observers argue the the technology could be used in cases of repression, few are calling for the outright ban Amazon cited in its statemement.
“Legislation is necessary to curtail certain uses of the technology and also fundamental misunderstandings as to how accurate the technology is,” said Garvie, whose work acknowledges that “police use of facial recognition is inevitable.”
“A lot of these systems have been put in place without formal notice to the public,” said Garvie. This means there has not yet been the type of public pressure and legislative scrutiny FRT deserves, she said.
Know a fact to enhance this story? You can edit it
Know a fact to enhance this story? You can edit itEdit