London to deploy live facial recognition to find wanted faces in a crowd
arstechnica.com
arstechnica.com
Although it catched a few fugitives - it triggered a lot of horror stories. A lot of false positives - which in an ideal world wouldn't be that bad. A police man would stop you, asking you a few questions and once he confirms you are not the guy they are looking for, they would leave you free. But it wasn't like this - even when the police man recognized you weren't the one looking for, they have to take you with them to the comisary, and spend a few hours here until a judge can set you free. Worst of all, there are a few stories of people being false recognized on a Sunday afternoon, and had to spend in the comisary until Monday when the Judge could go. Very very sad.
The current government recently placed Greenpeace, Extinction Rebellion and cycling advocacy group Critical Mass on a terrorist watch list (even though it admits they aren't terrorist groups). [1]
Previously we've had a situation where a cartel of 14 large construction firms operated a blacklist of workers with political opinions they disagreed with. Some were on the list merely for being members of trade unions.
This blacklist was compiled illegally with the help of UK police's Special Demonstration Squad. [2]
There seems to be a feeling that this sort of technology won't be misused but you can already see the signs that is exactly what will happen.
[1] https://www.independent.co.uk/news/uk/home-news/critical-mas...
[2] https://www.theguardian.com/uk-news/2015/feb/27/on-the-black...
No they didn’t. They placed their symbols alongside many others in a document which aids police forces in identifying both left and right wing groups of various persuasions [1].
Would you rather the police were ignorant of these groups and their associated symbols, such that they mistook peaceful protestors for terrorists?
[1] https://rebellion.earth/wp/wp-content/uploads/2020/01/Counte...
If it really was just a list of symbols for groups that might hold a peaceful but potentially disruptive protest then I would expect to see the Countryside Alliance or London Taxis Drivers Association on there... but they are not present.
When talking with people they sometimes say 'I did nothing wrong, this is just for the bad guys'. But what people fail to see is that 'bad guys' is a pretty fluid concept and it depends on the nature of the current government and who decides which people are on the lists.
Looking at the current British government it isn't even that hard to imagine the wrong hands that this technology could get into and cause damage.
Who maintains the databases with 'wanted faces'? Imagining a some corrupt corporation that just puts their enemies on these lists, think union leaders, critical journalists, etc, again, isn't that hard.
Never mind that this tech, being tech, isn't perfect and there are probably false positives that lead to very annoying situations.
Considering the dystopia 1984 predicted is a good idea, but one should also consider the failure mode of "Snow Crash" anarchic dystopia, where sufficiently-organized individuals use modern technology to harm innocent people and government cannot stop them. The mafia stories in the US stem from the era where city organized crime had firepower that out-gunned the police.
You are also missing something. It is really about civilians being harmed? How many people died last year from terrorism in the US or the UK? How many died because people were texting and driving? Or drunk driving? Should we track when you text and drive the same way? Should we have police sit outside of every bar testing people to prevent harm? I think we all prefer to have the presumption of innocence and to face consequences for our mistakes and crimes, not to be tracked in case we do.
> I want locating those members to be hard
I respect your opinion, but the people who have lost loved ones to "random" attacks don't generally share that view. Hell, part of the purpose of building a society and a government is to minimize that kind of personal violence against people.
I agree that government has a role in protecting people. We are both free to vote for representatives that are either for spying on citizens or against spying on citizens. We can also move if the other side wins.
I personally think the financial burden of retrofitting all US cars with breathalyzers is burdensome. If they already had such equipment, I'd 100% be in favor of gating starting the car on their use, especially if it was a completely passive, foolproof system. You're right on the number of deaths to drunk driving in the US; it's atrocious.
This wouldn't even be the first time in US history a policy change to improve safety and corresponding enforcement occurred; seatbelt laws exist, and enforcement of such laws allows the police to glance in your window to see if you're wearing one and fine you if you aren't. We aren't too far away from cars refusing to be driven if the seatbelt is unclipped.
To swing us back around from this analogous hypothetical to the UK's situation: the cameras are already sunk cost and the question on the table is to improve their utility via software. We aren't talking installing breathalyzers; we're talking about upgrading the existing breathalyzers' firmware to work better at the task they were intended for, so to speak. We can argue whether the software will technically work (separate question, and I'm not in favor of installing something that just doesn't work), but if it does work I'm in favor of passive mass surveillance of public spaces to find persons of interest. It bothers basically nobody and greatly improves the utility of a system that is already installed for the purpose of aiding in stopping crime.
As I’ve mentioned elsewhere, the police have admitted previously they’ve abused surveillance powers to blacklist innocent people from construction jobs simply for being members of trade unions. [1]
No body went to prison for that or lost their job and no governments were brought down because of it.
A few companies paid compensation but some are still fighting it in court.
Until we can point to examples members of the police or security service being in someway sanctioned for privacy abuses I don’t think anyone can say that the system is working.
[1] https://www.theguardian.com/uk-news/2019/mar/06/trade-unioni...
https://www.eff.org/deeplinks/2018/09/uk-surveillance-regime...
Until safeguards are proven to work and we have examples where they actually haven’t been applied then they effectively don’t exist.
(a) that the respondent State is to pay the applicants, within three months from the date on which the judgment becomes final in accordance with Article 44 § 2 of the Convention, the following amounts, to be converted into the currency of the respondent State at the rate applicable at the date of settlement:
(i) to the applicants in the first of the joined cases: EUR 150,000 (one hundred and fifty thousand euros), plus any tax that may be chargeable to the applicants, in respect of costs and expenses;
(ii) to the applicants in the second of the joined cases: EUR 35,000 (thirty-five thousand euros), plus any tax that may be chargeable to the applicants, in respect of costs and expenses; """
https://hudoc.echr.coe.int/eng#_Toc524359882
One can debate how effective a EUR 185,000 fine is in restraining the UK, but it strongly indicates the behavior was unacceptable and will be considered unacceptable moving forward. Seems safeguards work.
That cell harming people you want to imagine would be some other government agency acting as part of a covert operation after a long time of planning and preparation, with plenty of spies deep in the government. And they don't actually harm people all that much, not the point, they do just a few acts of terror with a few deaths for war propaganda. But once they are caught on surveillance it's already too late, it means they are ready to approach the head of police and start using serious military weapons, and plenty of people are already on the wrong side of war propaganda.
> where sufficiently-organized individuals use modern technology to harm innocent people and government cannot stop them
Such things don't really happen anymore, governments are too authoritarian already and really like their monopoly on harming people. The only way sufficiently-organized people can coexist with the government is for them to subvert law enforcement or to originate from law enforcement. Governments can deal with either case by replacing everyone or by letting them have their corruption in exchange for being loyal to the government.
Basically surveillance is not the tool for anything you talk about, it's role is the exact opposite: to keep people from disobeying the government, the ruling class, the whole systems, no matter how much harm they do to the people.
I don't judge the government for exploring facial recognition technology - there are motivating utilitarian arguments and they have the tools right in front of them. They are doing /something/, even if I think it's deeply wrong. But I do judge those who stick their head in the sand, or sit on their laurels, and that includes myself. We also have motivating arguments and the tools to express them.
I don't want to be part of generation of citizens whose children will say they learned nothing from history - beside the Nuremberg defense.
I'm in favour of basically every form of government snooping on an individual - whether that's watching your internet activity, tracking your car through ANPR cameras or spotting you with facial recognition - as long as you need to get a warrant to track someone.
The issue seems to be that the police can adopt (some of) these powers arbitrarily and with no due process (and in the past because intelligence agencies operate by a different set of rules). It would also provide an effective limit to the volume of people being tracked in this way, due to the overhead of acquiring a warrant.
Of course, there also needs to be a system to provide oversight and check that only legally authorised individuals are being tracked.
1) The government should not be able compel corporations to add 'features' to help them in this
2) We are talking about an actual warrant, where someone physically went in front of a physical judge int the same location to ask for it, and
3) There is no allowance for speculative but "safe" collection of data they might want under a warrant later - you can only access data occurring after you have the warrant, or naturally occurring records (i.e. no slurping up everyone's cell phone connections/conversations "just in case")
Broad based surveillance of the population by the government is just a bad idea. Cf much of human history on abuse of power.
Number (2) is to make sure this is an expensive process.
If should really be specifically enumerated with no slack. As in, surveillance footage keep for maximum of 1 hour (on line detection only, not historical movement search); only warranted faces will be produced in report, others people blurred.
Also, having unopposed, indefinite warrant applications is bound to be troublesome. Who is checking the validity of police statements? No one. If the police want the surveillance warrant to remain secret indefinitely there should be increasing levels of review. Likewise journalists, politicians, lawyers, medical data, etc.
In Canada I rarely see any evidence of surveillance outside of the cities. Public transit being the exception.
Weirdly, the one time I genuinely felt uncomfortable was last week. I saw a traffic camera* swivel across a junction. Actually seeing a camera move in such a way that it was clearly being operated by a human was genuinely unnerving. I took a wrong turn so it wouldn't see me, for no real reason.
*One of these things: https://d1ix0byejyn2u7.cloudfront.net/drive/images/uploads/p...
At least here in Northern Ireland the traffic cameras can be viewed by the public with either a live video feed or regularly updating images.
Maybe because, as most UK citizens (at least the ones I've spoke to about this), you know that most non traffic cameras are not under police control, but under the control of the establishment that installed them, bars, pubs, stadiums, office buildings etc and the police has to provide a good enough reason, via a warrant, to actually see the footage.
There is also the matter of logistics, to connect every camera in say, London, to a central location to process the data, you would need a crapload of infra that doesn't exists at the moment. Not to mention how much processing power you would need to use this tech for every camera.
I believe the most vocal voices to sort out the false positives will be the police themselves. Deploying the stormtroopers for false positives, a few times per day, is not going to sit well with them.
I knew a Mexican citizen who was a legal green card holder in the US. Unfortunately his name was a match to a fugitive on the most wanted list. Every US entry involved an hour long interview to confirm his identity.
This is what machine vision will to to everyone with its false positives.
That's definitely an impossible scenario, because it's not sustainable to interrogate every entrant into the country for an hour. So it's likelier that the ML algorithm will acquire bias to tone down the likelihood of false positive (at the tradeoff of occasional false negatives).
On paper the UK is living in some of the most authoritarian times in modern history. Post 9/11 is a completely different world across much of western society, we have less freedom in many respects. It's almost a trope to say we've given up freedom for the illusion of security. Quality of life isn't entirely represented by how many cars you can afford. I feel like we live in constant fear of terrorism, war, and so on. And yet we never actually get close to these perma-wars, they exist entirely in the press and on TV. I mean, how much Orwellian prophecy do you need before you recognize you are worse off than before?
Orwell assumed a system with a lot of power that was also unaccountable and heavily abused by those in control of it. UK government is neither of those things; they have elections more often than the US does.
I'm personally of the opinion that people take 1984 as prophetic gospel when there are multiple scenarios that include mass ubiquitous surveillance that could come to pass. David Brin looks through several alternatives in "The Transparent Society" that seem to reflect more closely the nature of the modern world.
Assuming the worst-case scenario is as much an anti-pattern for planning as assuming the best-case scenario.
From "The Transparent Society" Wiki page (I have not read the book)
> Brin argues that a core level of privacy—protecting our most intimate interactions—may be preserved,
The UK govt. has already invaded intimate interactions, to the point that they got hauled into a Human Rights court
https://www.eff.org/deeplinks/2018/09/uk-surveillance-regime...
I mean, how much more are you going to accept before they finally come for you?
Your example is the UK government being roped in by HUDOC, so I believe my answer is "Wake me when the UK isn't answerable to HUDOC."
In particular, the example you've cited is unauthorized and unjustified collection of private correspondence (intercepted communications), not collection of images of people in public and comparison of those images against persons of interest (a thing already legal to do by hand via human operators, which the computer algorithm merely speeds up). Private correspondence is private; why should people believe that their public movements should be so?
This is basically the 'finally they came for me' argument I was making. You will sleep in until it's too late is what you are suggesting.
> the example you've cited is unauthorized and unjustified collection of private correspondence
This was my point contradicting "The Transparent Society"'s optimistic POV.
I give up.
For example, HUDOC acts as oversight for the risk of overreach of the UK's police apparatus. Given that such oversight exists, the addition of facial recognition to do what the UK camera network is designed to do already (but faster and with programmable parameters) is probably acceptable. The benefits are clear and the risk mitigated by oversight.
[1] https://www.theguardian.com/technology/2020/jan/17/eu-eyes-t...
> London to deploy live facial recognition to find unwanted faces in a crowd
If they sell it like that I could see the tabloids foaming at the mouth with glee. :(
Pretty scary future we are walking into and it seems there is no way to stop it.
Also wondering if you can jam it by holding up printouts of people in front of cameras.
1. Use a VPN when connecting to public WiFi
2. DuckDuckGo for web searches
3. Use cash where possible
4. Billy Porter hat