AI cameras took over one small American town, now they’re everywhere | Hacker News Reader

AI cameras took over one small American town, now they’re everywhere

404media.co

55 pointswalterbell57 comments

Interesting to see how different countries approach this.

In Europe, the AI act is that it explicitly bans systems like these.

Two societies - one embraces such monitoring, another bans it. We'll have a nice experiment going on in the upcoming years :)

For certain "values" of nice... (says the person in the US, to be fair.)

France is "temporarily" allowing exactly this kind of systems in Paris for the Olympics.

How does the AI act bans "explicitly" this?

“If we have a guy that walks into a crowd with a gun and we pick him up on camera, our officers are on top of him in a matter of a minute, seconds, really,“

Why? It is not illegal to walk into a crowed with a gun in Mississippi.

https://gunlawsuits.org/gun-laws/mississippi/open-carry/ says "Mississippi is a permissive open carry state, and you do not need a permit to open carry within its territory."

https://en.wikipedia.org/wiki/Gun_laws_in_Mississippi says nothing about prohibitions related to crowds.

And even if it was illegal. At that point, the gun already shot people if it was there to be used. Cameras do not prevent crimes. Arguably, it should be very illegal to carry a gun (in a crowd) tho :D

> In part that outreach is because Fusus and its customers need town residents and businesses to hand over access to their camera feeds.

I wish there was more details on this. Did they create some kind of hardware that intercepts any camera at a low level, or does there need to be one device for nest cameras, another one for wyze, etc.

The article seems to imply it is hardware and not just software, but it's pretty vague.

The way I understand it is they provide hardware that lets them access local network-connected cameras (once given the necessary credentials), which is then passed on to the Fusus network by the hardware. Which is why they promise it works with every dumb camera.

So I shall give them access to my network? Not only my cameras but also my home network...

From the article it seems that residents & businesses need to buy hardware from Fusus: "I acknowledge that in order to participate in this program I will have to purchase equipment from FUSUS and pay any associated costs to access the technology required to transmit data."

So Fusus get paid twice: by people buying the cameras and by the police/authorities using the feeds from those cameras. There's good money to be made living in a paranoid surveillance society!

Unpopular opinion: Some time ago I decided to assume that everything I do outside of my house, and some of what I do inside of my house, is public knowledge which can be retrieved reasonably quickly. And if I'm to be targeted directly or indirectly (from appearance, faith, race, or some kind of behaviour or anything else) then I can defend against that.

I mean, to prevent or punish when an authority decides to abuse their power of observation (e.g. a stalker decides to follow someone via webcams and harass them), or to try our best to prevent or avoid the conditions that lead a nation state etc to target some groups (or individuals).

Also, if some authority decides to [do harmful thing] all people of x religion, it doesn't matter that much if they do it by AI webcams or "report your neighbour to the police" word of mouth.

Based on the experiences I've heard and read about, you're clearly trivializing the "defend myself" part and don't understand how problematic this is for the people actually affected by the discrimination or targeting you're talking about. It's not as easy as "oh, this is racism, you're in the wrong". These are institutions with comparatively unlimited resources with an interest in defending their behavior and how they treat the people they target. Lives are destroyed over this kind of thing.

So we should try to fix that, for us and everyone else. I know it's hard, but some things are hard. Preventing cameras etc is harder I think, because they can use "think of the children", except maybe in some parts of Europe it seems.

I think you have the right idea but both things can be right.

I think it's possible to simultaneously vote against "your worst enemies getting control of the state" and to agitate against the kind of widespread surveillance you wouldn't want your worst enemies to be in control of.

I do hear (what I think is) your argument that legislating against ubiquitous surveillance is counter to the inevitable tide of cameras and AI getting crazy cheap, a future where everything is a sensor and processing costs drop to zero. But I think (maybe naively) it's possible we can also have social norms and boundaries around how these things get used.

Sure, that's fair enough, I just think that they are winning so best prepare our expectations for it. We can still fight, too.

"think of the children" is partially a meme here in germany, there's a song about it, but it focuses mainly on how people wanna seem better by "thinking of the children" and less about using that as argument to negate human rights.

> or to try our best to prevent or avoid the conditions that lead a nation state etc to target some groups (or individuals)

So lets give govt (and a private company) all the power they want because we should just work hard so they don't abuse it?

No, that's a slippery slope. I mean, privacy is a lost battle, at least, outside your home, or on the internet.

You’ve clearly never pissed off someone with more money and connections than you.

My point is that if they would want to attack me or my community or things like that, especially illegally, or, in ways that would not really be approved by the public, then it would be easier to build up arguments or defences against those than to prevent cameras etc.

To make a comparison: it is also 'theoretically' easier to teach people not to shoot eachother, rather than to restrict their access to a technology which makes killing very simple. I think keeping the 'activation barrier' to bad actor induced, AI-aided stalking high is a good thing, even if it is only a temporary solution.

I think it could easily be harder to defend against an attacker who has the complete ability to surveil and track you at all times.

What about when they want to attack you for things that would be approved by the public? Imagine a black person in the Jim Crow south doing civil rights organizing having to live in a society with these cameras.

Let's say the public is against you (both in opinion and legally) and you need to organise your people to try to change their mind.

What I'm saying is that it's wise to assume that if you have enemies, they will know a lot about what you are doing and that if you do something in public or maybe even private, they may know what you are doing much more than they could in the past.

Is it right, should they do it? Of course not. But are they doing it? It's safer to assume, yes, they are.

Or walked past a random personality disordered obsessive who "fancies you" and decides to use their low-level digital privilege handed out to them as a petty council official or minor law enforcement role.

In fact, in some versions of experimental panoptic "utopias" [0], everybody gets to watch the feeds which are simply made public. Half your neighbourhood are in their homes watching the other half walking about scratching their arses and picking their noses.

[0] https://cybershow.uk/episodes.php?id=12

The point is to come to acceptance with the possibility that yeah, someone is watching me or you or our loved ones probably most of the time. If you are a celebrity you have no privacy anyway, and if you are a regular person you are too boring to watch, so, either way, not too much of a problem.

If the people in power have the personality disorders or tendencies to abuse their power then they hopefully won't be in office, because maybe someone else who watched them takes note of that, but sure, sometimes it happens. If they do abuse their power and get caught, then they should no longer have that power, eventually.

But this capability gives so much power to someone with authority to abuse it to remain in power…

That is correct, and it's wrong, but as a regular citizen, it's safer to assume that this is the world we live in now, and prepare accordingly, than to assume otherwise and get caught by surprise.

I cannot imagine what preparations can be made but since you have given it a thought, I'm very eager to hear. Maybe we can all benefit, who knows.

> I cannot imagine what preparations

I can, And it's very ugly and scary.

Go read the psychological literature on what happens when groups (of animals as well as humans) are put under sustained overt surveillance (not covert surveillance - they need to "feel" watched). Clue: it activates the "predators about" part of the limbic brain. Such a population is essentially a powder-keg in a pre-revolutionary state. For humans, on the surface it's all sweetness and smiles because social regulation (Freudian superego) keeps the lid on... for a while.

That part of the brain cannot be "trained out of it" so it cannot be "normalised". One cannot "make surveillance acceptable" because we have an innate aversion to it.

This is why in Bentham's design of the Panopticon the key feature of it is that nobody knows for sure they're being watched. Remove that limitation and surveillance is massively counterproductive and dangerous to everyone.

It's such a shame people don't read more about history, anthropology and human psychology before making parochial; statements about "the world we now live in" or what they think "people should accept". Before doing that it helps to research what _people_ really are, rather than go on personal gut instinct,

The British are still able to have their tea without losing their shit.

The psychological literature you are reading are written by people who are greatly motivated to produce such impactful conclusions. Life in reality is not that dramatic.

Aha, the old "Scientists are motivated to disagree with my worldview". Good luck dealing with your "reality".

Not all scientists. However we need to admit that quite a lot of them can be biased. Psychology has a serious reproducibility problem.

So... I follow Buddhism. The best thing is to be at peace with yourself, and accept that no matter what happens, everything is okay. Including death, loss of all material things, loss of sanity, imprisonment, and so on. This goes for yourself and also for your family. Then you become invincible. All of those things become only inconveniences. Sure, it'd not be nice to have them, so you can do something about them, especially your family or loved ones may feel sad more than you, because they may not think like you even if you may come to this kind of thinking.

I know that some people cannot do this and they think it's stupid or impossible, which is fair enough, it's too different from their upbringing and societal conditioning and so on. So let's say you still have chances to be devastated by a few of those things.

Now, let's continue with the assumption that "people in power" have the capacity to watch everyone, including each other. KGB of Chernobyl series style.

You can try to find out, as much as reasonable, what's the actual likelihood of your being affected by a particular thing you may worry about. And, decide, how much that would inconvenience you. Then, you try your reasonable best to reduce the chances of that happening, and also the impacts of it. If it's almost guaranteed to happen, for a moment, pretend that it already has happened. Let the emotions sink in, try to think beyond the emotions. What do you think would be the effects, to you and the people you care about? Can you do anything to reduce those effects? And so on.

It's impractical to do this for "everything that can happen", of course, so you can start with identifying your greatest worries and applying the process for them until your worries about those are reduced enough. And when other worries start surfacing more, apply the same process for them, too.

With more and more cameras everywhere we are coming to the end of the "who done it" investigative novels and movies and TV shows - at least those set in current times. Who needs a Sherlock with amazing logic and investigative skills when all the police and investigators need do is get recorded camera data. Police shows are becoming boring as they simply go from one camera to the next to discover the perp.

Yes, we have increased safety, and those intending to do us harm have less assurance of getting away with it, but something is lost (I'm not sure what it is exactly) in the knowledge that as we go about our lives we are being continually watched and there is less and less potential mystery and uncertainty in what's taking place around us and/or by us.

It would be more accurate to say that we're being recorded, and the recording is then viewed when something notable happens. Nobody is really watching you go about your business.

Almost everywhere where there are surveillance cameras there's someone who watches at least some of the feeds at a time for at least most of the day

The novels and shows will simply move the focus to the "who edited its footage" (okay already doing it - watch the Devs series)

Hm, isn't it possible to wear clothing that AI can't see?

So called ai is likely opencv and a model slapped on top. So to determine what it can and cant see you can play a bit with opencv to see how feature detection and extraction works.

Talking about feature extraction reminded me of these military exercises where they fooled a DARPA robot that was meant to be doing surveillance.

The military reached their target by moving inside a giant cardbox! Snake style!

It’s very model and hardware dependent

There's active infrared clothing which can partially protect you from cameras in general. E.g., a cap with 10 LEDs or so will probably hide your face most of the times.

That would probably overload the sensor so that your face would be less identifiable, but you would become instantly "that guy with the glowing face" which would make you much easier to track. Unless of course everyone starts using them, but that's another story.

I've wondered about doing this, just to make the point obvious that I don't like being face-tracked everywhere.

If I cared, I'd probably wear a couple near-infrared leds and blink them in a random timing pattern and combinations of lights so that cameras which pick up a little NIR would effectively be blinded.

aka the most easily trackable person

How to turn yourself into a beacon in 1 simple step.

Obligatory: https://xkcd.com/1105/

And if more than one person tries this, then:

> random timing pattern and combinations of lights

is an FFT away from being a unique tracking ID for that person! :).

There is no such thing. There were a few articles in press about such clothing, but it doesn't work. If a human can see you, an AI can just as well - with modern tech.

The only success was with highly controlled environments, and specific models - but these are extremely unlikely to scale to proper models.

I wouldn't expect most companies to use "proper" models at this point; some will, but for most (especially when used for mass-surveillance) I'd expect low computational cost models.

This will change over time, but just for the moment we do have an AI bubble exceeding the supply of people who can make good AI.

Even then such attacks only work under very controlled conditions and do not generalize across models. You might be able to fool one vendor/generation, but not all the others.

China is far "ahead" of any US city in this regard.

Reply on news.ycombinator.com