BigBrotherWatch: Misleading, incompetent and authoritarian: the Home Office’s defence of facial recognition
By BigBrotherWatch: Today, the Home Office finally buckled under the pressure of our Face Off campaign against lawless facial recognition and the government’s hoarding of 19 million photos of citizens.
Unfortunately, it didn’t resolve to make a meaningful change – but, instead, in a letter to Norman Lamb MP, issued misleading spin to veil an incompetent mess of authoritarian surveillance practices.
You can read the letter, from Baroness Williams (Minister of State for Countering Extremism) to Norman Lamb MP (Chair of the Science and Technology Committee) here.
Get Briefed, Get Weekly Intelligence Reports - Essential Weekend Reading - Safe Subscribe
This Home Office attempt to defend police practice comes four weeks before Big Brother Watch is set to launch a new investigative report, Face Off: The Lawless Growth of Facial Recognition in the UK, in Parliament.
We debunk the Minister’s six main defences below:
- “We will create a Board (to oversee facial recognition technology) including the three relevant regulators (the Biometrics Commissioner, Surveillance Camera Commissioner and Information Commissioner) and police representatives. This will (…) provide greater assurance that policing is complying with guidance”.
This translates as the Home Office vowing to plough ahead with automated facial recognition despite the lack of a legal basis; despite the concerns raised by rights and race equality groups; despite the clear affront to civil liberties; despite no operational case being made for its necessity; despite its cost to the public purse and astounding inaccuracy; and despite the existence of any relevant policy.
So when the Minister says this Board will “provide greater assurance that policing is complying with guidance”, what ‘guidance’ exactly does she mean? There is no formal guidance or policy on police use of facial recognition – and we would prefer that policing complies with our rights and respects the law.
We urge caution to the Commissioners – should they really agree to ‘oversee’ the exercise of a policing power that there is no legal basis or public consent for? In this legal and policy black hole, can they provide the right legal guidance, human rights analysis, and policy advice?
The creation of a Board merely provides the police and the Home Office with a fig leaf of ‘oversight’ veiling the lack of any legal basis for the deployment of automated facial recognition on our streets. The only thing this Board gives us “greater assurance” of is the Home Office’s dogged determination to roll out this authoritarian technology.
- The “’watch list’ may include those who are banned from attending the event, known criminals that operate in a crowded space or individuals on warrant for an arrest”.
Note the words “may include”. The Minister omits the fact that on Remembrance Sunday 2017, the Metropolitan Police used automated facial recognition to find so-called ‘fixated individuals’ – people not suspected of any crime, but who typically suffer mental health issues and have fallen through the care net. Function creep is already plain to see – and the technology is barely out of the box.
Despite this, the Minister writes that she has, “been informed” that at Remembrance Sunday “the watch list was populated with images of individuals forbidden from attending the events, as well as individuals wanted by police”. This is directly at odds with what the police officers operating the technology said both before and after they deployed facial recognition at Remembrance Sunday– that the watch list contained around 50 so-called ‘fixated individuals’ who were not wanted for any crime or arrest nor were forbidden from attending. Something doesn’t add up.
Nevertheless, the precedent that this justification sets is disturbing. The letter notes that in relation to Notting Hill Carnival, the “watch-list” of those wanted by police included “people involved in… sexual assault”. Why aren’t police using their resources to actively find the individuals wanted for arrest as soon as possible – particularly, in relation to offences as serious as sexual assault? Why are the police waiting for suspected sex offenders to attend public events and hoping they are picked out by their currently poor quality facial recognition cameras? Surely that allows a high risk to the public?
The idea that police should use hi-tech, intrusive surveillance and biometric tracking as a catch net for people wanted for arrest is absurd and is an approach that would inevitably rapidly securitise free public spaces.
- “AFR (automated facial recognition) is an evolution of ‘super recognisers’”.
It is plainly misleading to suggest that biometric tracking is an evolution of ‘super recognisers’ – officers trained to remember suspects’ faces and find them in crowd.
It is certainly true that the goal of each method is, at present, the same – both AFR and super-recognisers find people by analysing faces. But AFR technology is seismically different.
Individual officers don’t have the ability to biometrically scan every face in a crowd; to scan millions of people against watch lists of hundreds of people; or to automatically record biometric images of people walking by. Automated facial recognition significantly compromises the privacy of ordinary members of the public in a manner entirely different to super recognisers. Automated facial recognition creates biometric identity checkpoints on our streets.
Importantly, neither can super recogniser officers pose the risk of biometric surveillance and tracking at the population-scale that AFR does. Astonishingly, this risk is very real. The police force taking the national lead on deploying AFR has already openly expressed the desire to use AFR across the whole CCTV network.
We’ll be safe from that techno-totalitarianism for now – in fact, ‘super recogniser’ officers are far more accurate at recognising faces. But technology, especially AI technology like facial recognition, evolves fast – and this is a serious future risk.
Because if the Home Office takes the same draconian approach to automated facial recognition as it has with other surveillance methods, such as communications surveillance, then in a few years we may well find ourselves in a situation whereby the movements of all members of the public are tracked and recorded by facial recognition systems on the CCTV network. The prospect is chilling.
- “People are not arrested solely on the basis of matches made by facial recognition software”.
This is true – officers look at the alerts generated by the facial recognition system and decide whether to intervene. However, due to the extraordinary inaccuracy of facial recognition systems at present, police officers ask far more people to prove their identity due to incorrect facial recognition matches than they do genuine suspects.
What is especially concerning about this is the fact that multiple studies have found facial recognition algorithms to disproportionately misidentify black people and women. This may be due to the data on which the algorithms are trained – and police in the UK have refused to submit their facial recognition software for bias testing. This means that facial recognition could cause police to disproportionately require black and minority ethnic people to ‘show their papers’, increasing the over-policing of targeted communities.
Whilst people are not automatically arrested today, there is no doubt that it is a risk for the future. The Data Protection Bill, currently going through parliament, opens the door to law enforcement making purely automated decisions that affect people’s rights. We’re calling for a vital protection to be put in place that would prevent automated decisions being made where fundamental human rights are at stake. You can read more about our work on that issue here.
- “The images collected during these deployments are deleted if there is no match”
Our investigation reveals this statement to be simply untrue. We will publish further details in our forthcoming report.
- “Any weeding exercise (deleted innocent people’s photos from police databases) will have significant costs and be difficult to justify given the off-setting reductions forces would be required to find in order to fund it.”
Translation: the Home Office is refusing to give police forces the resources to delete data – images of innocent people – they unnecessarily and unlawfully hold.
There are 19 million photos on the Police National Database, mainly of individuals who have been arrested, many thousands of which are likely to be of innocent people. The High Court ruled in 2012 that the indefinite retention of these photos was unlawful – but the Home Office is refusing to take any serious measures to comply with the law.
A major problem is that, due to antiquated police systems, photos of innocent people can’t be automatically deleted – police forces have to manually delete photos. But because they haven’t been deleting photos for the past eight years (and they still don’t now), there is an extensive store of millions of people’s photos, including many of innocent people.
‘Costs’ are no reason not to comply with the law. And that shameful excuse is particularly facetious in light of some of the Home Office’s other spending priorities…
The Home Office awarded South Wales Police £2million to use automated facial recognition, checking people’s identities in real-time against those same custody images.
So if one thing is clear, it’s this – the Home Office is absolutely determined to roll out automated facial recognition in the UK – and it’s dragging its feet to avoid complying with law and respecting innocent people’s rights.
Big Brother Watch will fight this authoritarian move every step of the way.
Please donate now to grow our campaign against the police’s lawless use of facial recognition.