This is one of those instances where I say to myself, "I wish I had listened to Richard Stallman X years ago": https://stallman.org/amazon.html
Like Stallman, I'm so uncomfortable installing a listening device in my home like OK Google, Alexa, Siri, etc.
It creeps me out on a visceral level unlike any other kind of snooping.
Theoretically it's possible that the companies that build those devices lie and actually record you other times. But the same applies to any internet connected device that has a camera and/or microphone, such as pretty much every laptop and mobile phone.
The problem is that typical sloppy, glitchy software (and hardware) is poised to record your voice and send it somewhere at any time, and that somewhere will play or record this, because that's what all this stuff more-or-less setup to do. It's already happened! Alexas have already "butt dialed" other people (not just amazon's servers)!
Think a bit broader than that. Things can play out of various speakers in your house and influence these voice assistants. Again, advertisements have already triggered voice assistants. Research has shown how it is possible to play voice assistant commands that humans can't hear. These devices are poised and ready to record anything anytime and send it anywhere, or just use your credentials to do anything.
My laptop is different. It does have a webcam and a microphone, and straight-up locally installed malware could take advantage. But it's not quite the boulder-atop-cliff situation of the voice assistants. I've never enabled Siri, I've disabled it everywhere possible, I don't install junk software, I often check on things in Activity Manager because freakin web apps use so much memory/cpu/network these days, so I would notice. The webcam has a light which is hack-able but it's pretty hard. My phone has Bixby but I've never enabled it, I've never clicked "I agree", I've disabled the button and the swipe activation and the background services which I am able to without rooting the device. It's in reasonable shape.
It is a general problem that no one makes strong claims about indicators, so making a software driven camera led is not bad for a manufacturer's reputation.
Do you mind linking to the research you're thinking about?
Personally I physically disconnect all of the internal microphones from my laptops, and cover the cameras. I wish more people would do the same thing. Purism actually provides physical hardware switches which is much easier-to-use than having to physically take apart your machine (though Purism used to have proprietary kernel drivers, I'm not sure what the current status is).
How do you deal with the former on your smartphone?
RMS was smart to call it out at the beginning before it became more powerful.
HAH. Says who?? Are you seriously going to trust companies with a terrible track record and every incentive to do this, are you going to trust them not to use their always on listening device to listen to you? I swear I don't say this to derail the conversation but I literally cannot fathom in my head how you can think like that unless you're a paid shill.
These devices are also black boxes. Your laptop, if you're technically literate, is largely an open book as far as what it's up to, should you care to investigate. Granted, your phone less so, but it's also pretty easy to corral if you want to. The entire point of the always-listening assistant is that it's always on, listening. It would be awfully silly to buy one and then unplug it whenever you're not actively talking to it.
You used the word "Theoretically" like it's something so remote that's almost impossible which is both dangerously and hilariously not true.
Or at least: the ones with a smartphone. What makes this different?
For example: Trump is one of those people that can change the US into something you would not imagine a year ago.
It's very easy to say that this wont happen but I also thought Trump would never become a president...
The biggest lasting impact a president can have is appointment of judges. Congress pretty much has to do the rest, and they even have to approve the judicial appointments.
It typically doesn't happen overnight; rather, your rights erode over time, usually in the name of national security. That said, even in his short tenure, I believe Trump has done much to erode the norms of functioning democracy. Just a few examples: threatening political opponents with jail, threatening journalists, and at a very minimum having the appearance of basing foreign policy on how it affects his personal business empire.
Well... I for one couldn't imagine a primary channel for communicating American policy being a personal Twitter account.
My wife complains sometimes that I don't talk to her anymore. After quietly turning her phone off, and putting it in a distant room, I explain why ;)
The home pods are a bit of a drop in the bucket when you think about "smartphone coverage around the world".
What options are there?
"Of all sad words
Of ink and pen
The saddest are these:
'Stallman was right again'"
Not sure if this is the way to get people exited, building on better things.
I also like to read about, what works. What are working alternatives, and this he writes about ... but less so in my opinion.
Then maybe he is.
>He has the right to run his firm as he sees fit of course
No. If he is harming society then we the people should stop him, fuck his "right" to do what he pleases.
We're all equal when we are in public view, right?
Cameras facing all entrances and exits of the state capitol would be fed into ALPR and facial recognition software. There are enough photos of elected officials to train facial recognition. It would be easy to identify who is coming and going, what cars they own, who they travel with, and their patterns.
Public school was where children went to learn how to operate in public life. The rest of the population had less privilege, but more privacy.
Facial recognition services are incredibly close to neutral technologies in my opinion. If we truly want to stop abuse of them, it should be done with legislation. Getting mad and ranty at one of the many offerings (which also includes open source software) because it feels good to beat up a tech giant will have no useful impact.
You're entitled to that opinion. But it's just your opinion.
What I like to ask myself is: given that we know we had a corrupt individual in the highest office of the US government in the last 50 years (i.e. Nixon), how might that have turned out with each new technology that's exclusively under the executive branch's control?
I just think that positioning it as Amazon or Google or whoever doing a bad thing is misdirection. OPs article even seems to end up coming to a conclusion of needing policy to keep this under control. Someone will supply it to law enforcement, so it’s use more than its supply should be regulated because it does have myriad positive use cases also.
Building webcams that asses creditworthiness of store customers and the likelihood of returns is possible today. How long until it happens ? How long until we discover that badly dressed black people are often flagged by whatever tool comes out ?
Tracking people by their faces ? Been done, easy now (hell, opencv comes with it built in, so does YOLO). Tracking/identifying people by their movements (and I mean by how they walk, how they grab things, ...) ? Been done. Tracking people by the proportions of the limb lengths ? Been done. Taking a bad human tracker, feeding it into an LSTM and get vastly better performance when presenting it with video ? Been done.
And it gets much worse: outsmarting most of the human race ? Been done (let's be honest: humans have lost the Turing challenge. AIs are better at chatting up humans than other humans are). But no worries : humans still beat AIs at specific problems, and indeed humans still beat AIs ... well not at making AIs actually (genetic algorithms seem to beat everything else here). But humans still beat at "full system" AI architecture. For now.
And let's go further: how long until you can no longer talk to another human being ? Because that's exactly where voice recognition will lead. The 2030 version of Google Duplex : 
And of course, that makes the technology itself evil, not it's use by the executive branch per se. I'm sure they'll abuse it, but I guarantee they won't be the worst of your worries.
And it's MAD. Machine learning has joined the set of technologies that we must make sure we have. Otherwise, a conflict with, say, China is going to be thoroughly unpleasant (which would make them a lot more likely to start it as well). So we can't skip that.
And yet in this thread, nobody is talking about outlawing machine learning itself. And there certainly is no international effort to do that. Without that option, we might as well stop talking about measures we can take since they'll be absurdly ineffective.
You can "do no evil" but fact of the matter is since 1970 (and really since 1870) humans have been competing by robots. Humans can effectively no longer needlepoint, because they can't do so usefully, profitably, effectively, whatever you want to call it. Systematically the realm of humans is moving from everything to a niche.
Today, AI is a mostly-invisible salesperson making you purchase more stuff on Amazon.com, making you watch more advertisements on Microsoft/FB/Youtube/Google, and so on. It's literally choosing what to present to you in order to accomplish that, just like that friendly salesperson did when you walked into bed bath and beyond in 1995.
There is no solution here that is anywhere close to even being discussed.
For example, to give some vague indication of just how pervasive it already is: https://www.google.com/search?q=index+of+public+webcams
I guarantee you there's 5000 private webcams and non-web cams pointed at the street for everyone listed there. 100 government ones.
Stopping shopping is not an effective punishment, and will not lead to stopping progress on machine learning. I would just ignore it. Besides, at this point, it's probably just a security cam. 24h and your video is gone (assuming the store wasn't attacked).
Disclaimer: I work for AWS.
Do know what gives it away? The fact that Amazon is actively pursuing that market and isn't just "allowing anyone to use it freely and equally".
No, they're going to law enforcement and showing them Powerpoint presentations, demos, and everything.
With the exception of public figures, they aren’t sharing facial recognition datasets. This is a service that a moderately skilled team could implement with off the shelf and free software or an individual could do with a raspberry pi.
Law enforcement has been doing this for a long time in many areas. If there’s an ethical issue here, it’s with machine learning and camera surveillance in general, not Amazon.
Edit: especially if you run ads about how effective your guns are for serial killing.
By your standard, HN is on the same level as Amazon, as there have been many articles here over the years describing technical details on how to implement such a system available to anyone. A serial killer may be using a raspberry pi based facial recognition system that he learned about here, right now to stalk a victim.
That’s an untenable position.
Amazon's product takes care of all that nitty-gritty, thanks to the work of their many bright employees.
For implementing facial recognition, yeah, it takes a little bit of coding skill. Same as integrating with Rekognition.
"Amazon is providing police departments in Orlando, Fla., and Washington County, Ore., with powerful facial recognition technology."
Rekognition is SaaS (software as a service). Anyone can pay for running Rekoginition on their data sets (within Amazon ToS). Rekognition was not created directly for the police departments in FL or OR.
And that's whats happening to with a lot of surveillance technology, see Palantir, they are being packaged and aggressively marketed to schools, law enforcement and government.
Taken together these companies are selling and enabling a surveillance dystopia for no bigger reason than personal gain and the people responsible should be called out.
Sorry, pet peeve.
When this is used to track/recognize my face in a crowd what service is it that I'm receiving?
You can quickly identify well known people in your video and image libraries to catalog footage and photos for marketing, advertising, and media industry use cases.”
It’s even named with a 1984esque style.
Fixed that for you.
Now, there are real privacy implications to facial recognition and tracking. Serious ones. There are also benefits for law enforcement. Personally, I come down on the side that the privacy is worth more than the benefit to law enforcement. But that doesn't mean that that benefit isn't real.
Anonymity is one of the most effective tools for preserving free speech. If you can't speak anonymously, you're more likely to self censor what you say. If you can't consume information anonymously, you're unlikely to seek out or research unpopular views or ideas.
Barring massive social changes that are realistically never going to happen, a world where your identity and actions are always visible is a world without free speech.
Now, will this have a chilling effect on certain kinds of speech? Most likely, yes. Should we be concerned about that? Probably. But please, do not conflate 'free speech' with 'freedom from social consequences'.
Ignoring broader social issues and focusing purely on the government, citizen monitoring can be (and often is) utilized by governments to curtail free speech. People joke about the NSA monitoring them, but that absolutely has an impact on free speech - people are less likely to search something online the government finds distasteful when they know they're being monitored. Note that this is true even if the thing they'd search is legal.
If the government can monitor which events you show up at or what you read, does that make you more likely to be put on a no-fly list? Does that make it more likely for the government to pursue a separate legal action that it otherwise might ignore?
It has an impact on speech. It does not have an impact on freedom to speak. The fact that a social architecture changes behavior does not mean that it restricts it. This difference is important.
This does not mean that the right to anonymity is not important. It may well be. It just isn't the same thing as the right to free speech, and they shouldn't be conflated. I understand the desire to conflate them - anchoring the right to anonymity to the more recognized right of free speech gives it greater moral force. That's why people make arguments like that. But in so doing it lassoes the even more important right of actual free speech to the probably less important right of anonymous speech. It tethers them in a way that exposes the weaknesses of the lesser right to the greater one.
Free speech is the right to say whatever you want without government interference. The right to anonymity may be important to, but let it stand on its own.
> If the government can monitor which events you show up at or what you read, does that make you more likely to be put on a no-fly list? Does that make it more likely for the government to pursue a separate legal action that it otherwise might ignore?
Lots of things make you more or less likely to do various things. That is not the same as curtailing your right to do them. Changing organ donation from opt-in to opt-out increases the right at which people choose it. That is not the same thing as forcible organ harvesting.
Anonymity is a tool to shield people from otherwise unpreventable consequences. Separating them is like saying, "you have a right to privacy, not a right to encryption." One of these things allows you to get the other thing.
Think about why voting is usually anonymous. Most countries recognize that even with the general restrictions on government retaliating against speech, it is fundamentally impossible for us to block all of the many avenues it has to punish citizens. The only way to block that is to not allow it to know who specifically to target.
As a more modern example, America's current administration was willing to prosecute a bunch of inauguration protesters. If they had the ability to identify every single person at those protests, that would be a major deterrent to people attending them in the future. The government wouldn't even need to prosecute future participants; they'd just need to make sure that people kept in the back of their mind, "we could put you through a year-long legal battle if we wanted to."
Intimidation, targeting, and threatening is a form of suppression. We're not talking about marking a form opt-in or opt-out, we're talking about government interference into people's lives with the explicit goal of discouraging speech. It's not about whether or not people's speaking habits happen to change, it's about why they change - because they're frightened of government power to punish them for searching or expressing unpopular opinions.
People aren't conflating anonymity with free speech because it makes the argument stronger, or because they're trying to sneak it in on the side. I couldn't care less about anonymity, in except that it is the only strategy that anyone has ever come up with that allows me to have a large number of other fundamental rights that I do care about.
The same is true of speech. It is a right that exists to protect other rights. Speaking, in and of itself is nice, but the function that right serves is the circumscription of state power.
> Anonymity is a tool to shield people from otherwise unpreventable consequences. Separating them is like saying, "you have a right to privacy, not a right to encryption." One of these things allows you to get the other thing.
I totally agree. That doesn't make it equivalent to the right to speak, however. The fact that anonymity augments the right to speak is great. That may be an argument for having a right to anonymity. It does not make the right to anonymity equal to the right to speak.
> Think about why voting is usually anonymous. Most countries recognize that even with the general restrictions on government retaliating against speech, it is fundamentally impossible for us to block all of the many avenues it has to punish citizens. The only way to block that is to not allow it to know who specifically to target.
Again, totally agree. We don't disagree that attribution has a chilling effect on speech. You're preaching to the choir here. We disagree that the right to speak is inalienable from the right to have your speech unattributed.
> People aren't conflating anonymity with free speech because it makes the argument stronger, or because they're trying to sneak it in on the side. I couldn't care less about anonymity, in except that it is the only strategy that anyone has ever come up with that allows me to have a large number of other fundamental rights that I do care about.
Your argument seems to be that "anonymity makes the right to free speech stronger, therefore the right to speak and the right to speak anonymously are equivalent". Which, to my reading, does not follow.
I think there are great arguments for the right to speak anonymously. I think you have articulated several of them nicely. I personally believe in the right to speak anonymously. But it should not be discussed as if it is the same right as the right to speech, because it is not.
Anonymous speech has costs and it has benefits. One of those costs is president. I still think it's worth it on balance, but if you say that anonymity and speech are the same, you're saying that arguments against anonymity apply to speech, and that's a dangerous game that I, and I don't think even you, really want to play in this political climate.
Anonymity is an implementation detail. At the moment, it is an implementation detail that is essential. If your point is that it's not literally equivalent to free speech, then sure, definitely. We're on the same page and I agree with you.
But for all its warts and problems, if anyone wants to get rid of anonymity, then it's their job to come up with an alternative implementation. Part of supporting free speech is figuring out how to practically guarantee it in the real world, the rest is just wishful thinking.
> Speaking, in and of itself is nice, but the function that right serves is the circumscription of state power.
Quick side note, but this is a very narrow reading of free speech that I expect many people would disagree with. The function of free speech is to facilitate a person's fundamental right to share and explore ideas. Limiting the power of the government is a means to that end, not the other way around.
The idea behind a fundamental right is that it is... well, fundamental. Government doesn't give it to you, it respects that you already have it. Imagine how silly it would sound to say that freedom of religion, or racial equality, or the right to discover your own gender identity existed just so we could keep a government running smoothly :)
I used to agree with this religiously. But now i'm not so sure. Anonymous speech is great for countering authoritarian regimes. It's great in China, Russia, North Korea. But those countries don't have freedom of speech in the first place. I'm all for anonymity technology there.
But in places that do have freedom of speech, what does anonymous speech really do for you? It seems to me that it mostly induces polarization. It encourages a certain kind of Overton-window stretching that has so far proved mostly toxic. I can't think of any examples to the contrary at the moment, though i'm sure a few must exist.
> Quick side note, but this is a very narrow reading of free speech that I expect many people would disagree with. The function of free speech is to facilitate a person's fundamental right to share and explore ideas. Limiting the power of the government is a means to that end, not the other way around.
Ok, yes, in a certain sense that is true. I don't mean "the reason people ought to be allowed to speak". But moreso, "the reason we ought to enshrine free speech in a constitution, making it especially difficult for even our democratic government to alter or amend". I think the reason you want the latter is because in a democracy, an informed public is essential. If the government has the power to limit speech, they have the power to reshape voting behavior in their own image, which short-circuits the meaning of democracy.
For everything there is a use that could enable any dystopia, doesn't mean that it will actually enable it.
If AWS is "enabling dystopia", so are Linux and Ethernet.
You can do positive things with Rekognition. You can do dystopian things with it, too. Get AWS to stop doing it and folks will just use GCP and Azure's similar offerings.
This is a problem that fundamentally needs government regulation to be successfully addressed.
This thread seems to be following the general logic that because Amazon is a big company that uses AI/ML, they must be using it for evil. I’ve never seen a thread on HN that accuses OpenCV of facilitating dystopia, or compares the project to selling guns to a serial killer.
The NYT (and C-SPAN, apparently) uses it to identify members of Congress:
Shutterfly is apparently using it for image categorization:
I uploaded about 120k photos including all my father's photos I had scanned. (about 4000)
I injured my toe when I was 10yrs old. Searched in my Google photos "toes". Found the picture pretty quickly. Have had similar success finding pictures of my childhood dogs. Can also search by people (face recognition).
It's still got a long way to go but I love it when it works.
Clarifying edit: I'm not being disingenuous. I may be too close to the technology to grok why this thread is a down-vote minefield. Running OpenCV on an Arduino board is years-old proof-of-concept stuff and now I have an AWS DeepLens sitting on my desk that can name which of my neighbors walked their dog past my house.
> In order to protect civil liberties without hampering innovation, lawmakers should require public input before surveillance tools are deployed and ensure that facial recognition databases are purged of data related to law-abiding people.
> lawmakers should ensure that real-time facial recognition capability is not merged with police cameras, whether they are dash cams, CCTV cameras, or body cameras. Facial recognition may well be a valuable investigatory tool, but outfitting police with real-time facial recognition capability will only increase the likelihood of needlessly contentious and violent confrontations between police officers and members of communities across the country.
> lawmakers should require public input before surveillance tools are deployed
Again, what does that mean? Input is not the same thing as saying the public can veto it. Even if it does required explicit approval from the public is that really a great solution? The public has made a lot of shitty political decisions recently. Furthermore, who counts as part of the public? If you're a convicted felon who served his time do you count as part of the public?
This shit reeks.
Totalitarianism, forever. Take solace, though, citizen: it's not just the government tracking everything you do, it's everyone!
It makes me sick. I wish I had never gone into CS. I too am a collaborator. If it were up to me, we'd destroy the machines in a Butlerian Jihad. The road we're going down is dystopia with no way out. The tools of control will be total, all-encompassing, omniscient.
Philosophically I would rather laws and social norms are enforced by law and not corporations or individuals with disproportionate influence.
My comment is clearly about the danger of imposing morality through corporate influence.
I did not say that a person’s entire responsibility is to enrich themselves. I said that is the responsibility of an employee. People can and do have complex and sometimes conflicting interests and responsibilities.
Someone needs to design an AI-powered camera-seeking spraypaint drone.
Second, it seems to be a popular misunderstanding, that GDPR affects Europeans everywhere they go and you need person's confirmation to store his data. It affects people currently located in Europe, and there are listed multiple very vague excuses for storing user's data without explicit consent, like "in order to protect the vital interests of the data subject or of another natural person", with which global tracking by law enforcement agencies could be easily justified.
On a more serious note; stuff like the US amendments is often already included in the base law (the generic name for constitution) as they are much older than the US. The times where the base law was somewhat more moldable (like before widespread literacy) made that possible. For example, freedom of religion is pretty universal, except in countries where a ruling religion exists which inherently to the religion bans all others. Arms are usually removed from base laws a few hundred years in as they directly conflict with the laws that ban killing people.
And all in the name of combating terrorism.
This issue is political in the sense that it's very much about social policy... and it's just possible that it's bad social policy to let people perform/offer surveillance-at-scale as a service.
(It is kindof funny to see Cato discover one significant limit of a general philosophy of private market freedoms, though.)
(Or, for that matter, Amazon.)
Regardless, the issue is real and concerning.
Even worse, knowing about this phenomenon seems to do very little to stop it. At least that's what I'm feeling.
All it takes you to do the same is create an AWS account, fill in some info so they can bill you for the services you use, a bit of code and some camera's.
I know it's very popular these days to quickly blame it on the big tech names, and yes they are doing a lot of shady things, but articles like this are simply name calling and not explaining how things really work.
Besides that, this technology is not new. Xovis camera's for example have this ability build in for a long time already. Those camera's are installed in a lot of public places like shopping malls, railway stations, airports etc and they are used to track you. Just because Amazon made a scalable central version of this doesn't make them track you.
People who write articles like this should do some research first ...
But IMO no amount of hand wringing will stop this inevitable total surveillance short of a constitutional amendment.
I imagine a high-profile incident with privacy violations will turn up one day and make the public interested. By the time that happens it would be beneficial to already have a privacy amendment proposal.
Paparazzi and autograph hunters, to name two obvious examples.
Once a thousand startups are using this in a non-secretive way, maybe voters will wake up and demand regulation?
It may be a side-effect, but Kudos to Amazon regardless if it does raise awareness.
EDIT: fixed a spelling error
Likewise, we're sitting here focused on Amazon for implementing facial recognition instead of focusing on changing laws and government "culture" to make such an act against its own citizens far from accepted. I don't like what Amazon is doing either, but attempting to stop people from advancing facial recognition seems fruitless at best. Facial recognition is here, and it's only going to get easier, better and faster.
Can we focus on finding ways to prevent/identify/etc bad actors? Because facial recognition is coming, regardless of Amazon, and bad actors aren't loyal to amazon.
Tech like this is designed and deployed in such a way as to discourage people from thinking it's a problem that needs a fundamental solution. If people realize it's a problem at all they assume it will be solved with just a few laws or a company protest.
If there was a public website you could go to where anyone without authentication could monitor every street in your town, maybe that would spark more citizen-controlled solutions, in the same way that browser extensions intercepting Facebook logins at coffee shops sparked https adoption.
The problem of democratizing surveillance is that many laws prevent neutral actors from building these systems (what's your state's 2-party consent law for audio and video in public spaces?) As a result, the only people doing widescale recording are exempt institutions and companies that people are trained to think aren't a problem, or criminals that people think are rare or limited because their results aren't widely broadcast.
It's like social security numbers. If a company started just publicly releasing them en mass via its Twitter account, we would quickly figure out an alternative way to authenticate that prevented them from doing that. But if instead a company just leaks them to the black market, well that's just a security breach and the system doesn't need to change - even though the practical effects are very similar.
It's trivial to secretly monitor people via facial recognition. It's not trivial to publicize or demonstrate how easy it is in a dramatic way that will spark public attention.
Amazon marketed it to police explicitly, I'm pretty sure the open source team working on TensorFlow isn't going around suggesting you plop a model on Linode and use it to catch criminals. This isn't something that police departments have the ability or skill to discover and implement on their own, setting the menial bits up for them and suggesting a solution is half the battle with government stuff like this.
The Seattle Police Department now has a division focusing on data analysis. This came out of a consent decree with the Justice Department to study compliance for use-of-force incidents, but is now being used for other things to quantify how well the police are using resources in different neighborhoods. Recently they sponsored a hackathon sprint for local developers to come up with ways to quickly anonymize police body cam footage so that citizens who were not the subject of an investigation could not be identified, but still provide requested footage in a timely manner.
I don't see that in the article. Can you provide a source for that?
> This isn't something that police departments have the ability or skill to discover and implement on their own, setting the menial bits up for them and suggesting a solution is half the battle with government stuff like this.
It's something I have the ability to do on my own. Any police department could hire someone like me.
If you ever have children, wait for them to make this same argument. It should happen around the age of five.
Because some people respond to a hypothetical better when they can see concrete aspects of it, not just hear a description of the possibility space.
"I could build that if I wanted to" vs. Recognition is a bit like the difference between the black market and dispensaries, a difference of scale is a difference of kind.
It's a band-aid, but band-aids are an extremely useful medical tool. They only become problematic if used as an excuse to avoid stitching the wound.
Sure makes you think...