Facial Recognition - Friend or Foe
Should we be worried that everything from smartphones to the tax office knows what we look like?
Last week, the US IRS (tax office) backtracked on a decision to force all American taxpayers to use a third-party facial recognition service to file their tax returns. In 2020, the South Wales Police were taken to court over their use of Live Facial Recognition. The Met Police also controversially use the technology to scan for criminals, and, last year, nine Scottish schools started - and then quickly stopped - using it to verify children's identities.
Despite these seeming setbacks, the technology is widespread and growing. For years, airport security have employed facial verification to protect borders, and phones and laptops are often unlocked by just looking at them. It's certainly much more convenient than having to type in a password.
But there is a fundamental difference between these different types of technologies, as Stephanie Hare explains in her new book, Technology is not neutral. In the broadests strokes, they fall into two camps: facial verification, and facial recognition, which she terms 1:1 and 1:many.
Facial verification is concerned with figuring out if you are who you say you are (think unlocking your phone), while facial recognition tries to answer who you are. Both are contriversial, though for different reasons.
Let me see some ID
The US Treasury recently announced that to combat tax fraud, facial identification will be required to log onto their online systems. They have since reconsidered. The third-party involved, ID.me has been involved in controversy over whether it uses 1:1 or 1:many technology and its use of Rekognition, a system that Amazon no longer sells to law enforcement.
Last year, a number of Scottish schools started and then stopped requiring facial identification from children to pay for their canteen meals. Fears were raised over privacy and normalising the practice.
Nonetheless, many iPhones and laptops are unlocked using facial recognition on an hourly basis and border security uses the technology in eGates. As Hare explains, we could consider these as low-risk. "Its just you and your device... The biometric never leaves your phone."
This aspect of control and consent is key to the distinction between facial verification, facial recognition and other biometrics. According to Gareth Mitchell, presenter of the BBC's Digital Planet, "If somebody says to me... can you just identify yourself using your fingerprint, I can make a decision whether to put my fingerprint into that reader or not." We are given the choice to use our faces to unlock our devices.
When that choice is eliminated, as in the Treasury or canteen examples, disatisfaction mounts. And we don't have a choice whether or not to wear our face on the street, which is why facial recognition keeps stirring controversy.
Halt! Who are you?
If you walked through Oxford Circus in London on the day the US Treasury backtracked, your face may have been scanned by the Met. Over 12,000 people's faces were compared against a watchlist of 9,756 resulting in four arrests. Only one person was falsly confronted.
This contrasts heavily to the previous deployment in 2020. There the system made seven false alerts out of a total of eight. Six people were engaged by the police and one arrested. The Met claims a false positive rate of 0.08% (seven out of 8,600).
This kind of Live Facial Recognition, or LFR, is the most controversial. Police claim it helps them scan large numbers of people quickly and accurately for fugitives. Critics point to its sketchy track record with identifying people of colour and women. These are some of the reasons Amazon halted providing Rekognition technology to law enforcement following the George Floyd protests.
Also last week, Lord Clement-Jones opened a debate in the House of Lords on the topic. He claims "there are numerous other threats to human rights that the use of facial recognition technology poses. To the extent that it involves indiscriminately scanning, mapping and checking the identity of every person within the camera’s range—using their deeply sensitive biometric data—LFR is an enormous interference with the right to privacy..."
Time to face the music
One thing is clear: we need to have more robust conversation around this technology. Repeated rolling-out and retraction of the practice of LFR doesn't help anyone, and there are other techniques such as Retroactive Facial Recognition that need frameworks.
Its a difficult discussion, but the question of privacy versus security might well be one that shapes our age. And as technology grows ever deeper into our daily lives, we need to find some answers.