Yesterday the FBI signed its first public contract with Clearview AI | Hacker News Reader

Yesterday the FBI signed its first public contract with Clearview AI

twitter.com

241 pointsdanso138 comments

So Clearview has a database of billions of faces scraped from the internet, and sells subscriptions to match photos on that data. I don't think we can put this capability back in the box. There is no moat around it.

Keyhole once had a similar business model for satellite data. Then Google bought them and offered it to everyone for free, making a fortune. I hope that Clearview's path goes the same way. If this is available to law enforcement and anyone else who can afford it, better for it to be available to everyone.

If big brother can have it, little brother should too. There is no serious prospect of withholding it from big brother. I think I'd rather have privacy from both, but that doesn't seem to be an option.

The "moat" in this case would be overwhelming legislative action, criminalizing Clearview and its ilk out of existence. The law (supposedly) exists to ensure and protect the commonweal, and we should apply it instead of sitting on our hands and lamenting the market's unwillingness to self-legislate.

That probably won't happen, of course.

this only works in one country at a time, you can rest assured that other governments have your face tucked away in a datacenter

Sure. I'll work on fixing the place I live, sleep, and eat for the time being. Liechtenstein can wait!

But can the other Five Eyes states wait? Offshoring outlawed domestic surveillance to allies seems to be the name of the game now.

I'm not confident that anything will stop US military and defense intelligence (which is completely distinct from the FBI, as a domestic LEO) from doing shady things.

Instead of worrying about how they're going to get around the law (they can do parallel construction on the Moon, for all I care), I'd prefer to have a sufficiently big punishment waiting for them when the public finally learns about it.

Time will punish them. As it did with any imperium in the past, as it will until the last ape on this space rock has ushered their last breadth.

Time is lord of all of us. Peasants and emperors alike

There is no evidence that this happens, and it is illegal for the US government to ask another country to spy on its citizens. If it happened, the resulting lawsuits would make headlines for weeks, and there would be forced resignations.

> There is no evidence that this happens ...

Didn't the Snowden documents and various things from that point give evidence that it _does_ happen?

GP found HN that year, might have missed the headlines.

https://en.m.wikipedia.org/wiki/Global_surveillance_disclosu...

Point me to any document that says the US asks other countries to spy on its citizens.

No. Nothing in Snowden's leaks showed this happening. If anything did, there would have been lawsuits. People believe there were a lot of things in those leaks that weren't actually there.

Other governments lack the ability to take my property, take my freedom, or take my life.

FBI has the ability, with a single action, to ruin any resident's of the US life, I think that is significantly more worrisome than if Russia or China has my face unless I lived in Russia or China

I didn't down-vote you as I believe views like this should be stated but responded to properly. Your safety from foreign governments is not guaranteed even if you live within the borders of the United States. A perfect example of this is a post by someone else here on Hacker News pertaining to the tools that the Chinese government uses to identify dissidents outside of their borders.

https://www.nytimes.com/2021/12/31/business/china-internet-p...

Nothing in that story seems to concern me or most US Citizens, if you are a Chinese citizen living in the US, attended a US University then return home sure you should be concerned and I do not condone the actions of China. However I am a Natural Born US Citizen, and I have no plans or desire to ever visit China, Russia, or even Europe

And lets be real here. FBI is not much different, and has more direct control over people in the US than the china bogey man you are attempting to peddle

I'll preface this with the fact that I'm very pro-privacy. I don't so much object to certain technologies or data collection; the cat is out of the bag to some extent. I do object to the conspicuous lack of regulatory frameworks. For instance, if you're going to use Clearview on a suspected terrorist my first question would be, "What due diligence and groundwork have you done to prove that they are in fact a terrorist?" If it's a known member of the Taliban who has a Twitter account that regularly calls for death to Americans, is that enough -- or is that some kind of edgy free speech? My point being, I'd like a very high bar to use these technologies and I'd like for them to be, mostly, used in apprehension rather than intelligence gathering.

I would suspect in such cases they would have a 'case file' on such an individual. Perhaps it ought to be to the extent it would pass any reasonable jury in a court of law - I would be very surprised if this system isn't merely an addition to their existing toolset to help them validate/process data (perhaps as a verification system? finding individuals they are looking for?)

Here in Australia we are continually passing laws removing judicial/court/judge oversight for things and replacing it with whimsical oversight you’re lucky if someone outside of the requesting organisation looks at it. It’s pissing me off :(

Australia's rapid dismantling of any checks and balances is extremely concerning. It's also something of a bipartisan stance, the only real pushback tends to come from independents and minor parties.

> "What due diligence and groundwork have you done to prove that they are in fact a terrorist?"

The solution is to make this question their (Clearview) problem. If you are offering a service that makes damaging claims about people, then you need to be liable for any related damages if that claim is later found to be slander/libel.

The risk of a potentially huge ruling/settlement will be handled the same way it is handled in other professions: by paying for liability/malpractice insurance. Eventually, the insurance companies will handle the question of due diligence.

Unless you make a new liability law that is pretty specific to this use case, I don't think it's very possible to make slander/libel damages for it, especially considering the US's 1st amendment laws.

Clear view probably today says "I think this is %90 likely to be person X, but I could be wrong" in their UI. That turns it into opinion.

The regulation for this is not libel.

The cat is out of the bag? So unleash the dogs!

"Cry havoc and let slip the dogs of war"

That is why classified systems exist, they act as a last ditch mechanism to ensure that the ruling class isn't inconvenienced by silly things like "the law". How many people have been locked up for lying to Congress about their illegal activities at the NSA? What about the people at the FBI who knowingly misrepresented evidence while renewing secret FISA court surveillance orders? You remember the so called "Satanic Panic" in the late 80's? Did you know that documents were declassified last year proving that it wasn't a silly conservative overreaction to Iron Maiden, but something that was actually happening - and the CIA was (for whatever reason) coordinating the media's handling of events? Do you think the law hindered those government employees assisting in the trafficking of those children, or saw to their punishment? Nope.

> You remember the so called "Satanic Panic" in the late 80's? Did you know that documents were declassified last year proving that it wasn't a silly conservative overreaction to Iron Maiden, but something that was actually happening

[citation needed]

These are the keywords you are interested in: The Finders, FBI, CIA memo to NBC, North Korea visa. It has been my experience that the more effort one puts into sharing something like this with a stranger on the internet - the more easily it is dismissed. You've got a few hundred pages of declassified documents waiting for you, do the work or don't - it really depends on the sincerity of your "citation needed".

I honestly have no idea how truthful any of this is. If experience has taught me anything, it's that conspiracy theories like this are largely bullshit. Thank you for at least providing me somewhere to start though.

huh, you'd think that the last few years would have sufficiently chastened the kind of people who reflexively characterize things as "conspiracy theories"... :)

No, not really.

Yeah, you're right - it was definitely bat soup and Adam Schiff will be releasing that evidence of watersport hijinks any day now.

> That is why classified systems exist

This is historically as well as prima facie inaccurate. Yes, classification is abused. No, that is not why it was created nor why it is maintained.

Why it was created? I don't think anyone could offer more than guess. Why it is maintained? lol. First, we could get a lot more specific about the kinds of classification and the distinguishing characteristics... but I think we can get away with this: broadly speaking there are two kinds - tactical and strategic. Tactical is almost always SECRET (sometimes with a further qualifier like NOFOR) and below. This is stuff that is very time sensitive. For example: Manning's SIPRNET data dump. It makes sense to protect this information for a short period of time, no more than 2 years - as declassification would only skyline bad actors. Strategic (everything above SECRET) is where are all the highly illegal stuff that would cause a massive amount of backlash is hidden. For example: all of the Snowden/Greenwald releases. Sure, there are some long running strategic programs that could be argued needs protection... but to a degree that justifies the entire security apparatus? No, that is the whole reason why nukes are suited to their purpose - the barrier to entry is far higher than getting your hands on some schematics.

Clandestine sources always pale in comparison to open source collection, when the objective is the truth - not some concealed agenda. So, the idea that the status quo is maintained for any other reason than concealing bad behavior is laughable - we know that it doesn't do what it purports to do.

Yes, this.

Legislation and regulation operate on the basis of finding some viable point of control or enforcement.

The devices are ... ubiquitous. Collection all but certainly cannot be stopped.

What can be controlled is:

- The collection, solicitation, retention, sale, purchase, or transacting of the images, data, or access to each.

- Voiding of any contractual obligations concerning same. Any such business would literally be outside the law.

- Development of technologies or training sets concerning such data.

- Validity of any such data or conclusions derived from it in courts of law.

- Any business decisionmaking based on such data. Protected by whistleblower laws.

As a start.

Standalone collection, security footage recordings, and the like might be permitted with some limited retention period (say, 1--3 months), but restricting any aggregation of such data. Individual images could be taken. Posting identifiable images online without the freely-given willing consent of subjects should be sharply limited. Yes, that includes children. Compelling public-interest exceptions might exist.

The biggest challenge will be totalitarian surveillance states which practice and cultivate such surveillance, and which drive further technology and practices both within and beyond their own borders. That is a problem whcih inherently goes beyond mere law, and always has.

Privacy is the ability to express and enforce limitations on access to and distribution of information of, by, or about you. It is an emergent principle, and its own scope and definition expand precisely as information technologies do. In an age of speech, it concerns gossip, hearsay, and slander. In an age of print, libel. It has expanded with photography, telephony, improved optics and sensing (long lenses, infrared imaging, microwave scanning of structures), and with ever-expanding data storage, processing, and distribution.

In response to an earlier discussion where it was asserted that concerns over privacy in information technology are somehow a post-1980s phenomenon,[1] I compiled a list of notable persons in the field who'd voiced concern earlier, one of the most notable being Paul Baran, a co-inventor of packet-switched networks whilst working at RAND in the 1960s. At my request, his RAND monographs are now freely accessible to the public:

Paul Baran:

- "On the Engineer's Responsibility in Protecting Privacy"

- "On the Future Computer Era: Modification of the American Character and the Role of the Engineer, or, A Little Caution in the Haste to Number"

- "The Coming Computer Utility -- Laissez-Faire, Licensing, or Regulation?"

- "Remarks on the Question of Privacy Raised by the Automation of Mental Health Records"

- "Some Caveats on the Contribution of Technology to Law Enforcement"

Largely written/published 1967--1969.

https://www.rand.org/pubs/authors/b/baran_paul.html

I've listed additional authors and references here:

https://toot.cat/@dredmorbius/105074933053020193

________________________________

Notes:

1. See: https://news.ycombinator.com/item?id=24745246

At the end of the day, our computers and smartphones are just data input/output machines. Humans generate petabytes of data yearly, and that data is going to be stored, processed and used, forever. The invent of HDDs, fibre internet and increased processing power means more can be done with that data. It's entirely plausible entire streams of stored encrypted data (TLS, openvpn etc) may one day be decoded and provide fascinating insight's into human behavior. (I wouldn't be surprised if there was a giant archive of every single byte of FB/Twitter/Reddit/HN data in a DC somewhere...even this post!)

I think we need to accept and embrace this, and that privacy is now a fight that is, for better or worse, a lost cause. Covid generated even more streams of data through sequencing, and you can be sure all of this is being stored, forever.

You can, and probably should, fight against it, but if you're a human alive today you should know many bits of data (about you) will almost certainly be contained on many disks for future generations to look at, and that isn't likely to change any time soon (short of a major regression in human capability and understanding)

Has anyone tried suing Clearview for copyright infringement? It seems like wouldn't have a legal right to be using these images.

I would guess they would cite transformative use.

If they are making false claims about you that cause serious damage to your reputation (like telling the FBI you look like a terrorist), then sue them for libel.

Wouldn't you have to prove you don't look like a terrorist?

Clearview is probably saying that if you convert an image, or set of images, of a person to vector(s) then there are some similar vectors that are associated with a terrorist. If they are saying it, it's because it's probably true in their implementation.

I probably do look like a terrorist somewhere. The problem isn't identifying that person and associating them with me, the problem is treating that information as significant, and that's on the FBI (in your hypothetical scenario).

I think that one potential solution would be for the SCOTUS (who has made very wrong decisions in the past) to rule that third-party information is protected and requires a warrant just like entering your home. The decision to allow warrantless requests from businesses was a major error (or, maybe, malicious decision) in my view.

Step 2 is to ban such activity or rule that it is copyright infringement. And to maybe admit that scraping needs more guardrails and regulation.

Maybe we should just leverage the existing copyright penalties our beloved MPAA and RIAA lobbied so hard for? What is it... 15 years in jail and $250,000 fine per violation? Isn't that what the scare screens on DVDs always warned me about? If they're claiming 1B photos they've scraped and violated copyright on...

Rule #1 (as famously learned by SpaceX and Tesla) about doing new and risky things: Don't ask permission. Deal with the consequences later. They'll probably be smaller than if you had asked.

Billions of faces sure. How many of those are duplicates? How many of each person do they have and in what quality? A low res snap from a traffic camera is useless.

To distinguish individuals you need a lot of data.

"and offered it to everyone for free, making a fortune" - this makes no sense

This is Google Earth and Maps. They're free to users, but make Google billions on ads.

> This is Google Earth and Maps. They're free to users, but make Google billions on ads

Citation needed. Google only recently started showing ads in Maps and there is no way it's near billions at this point.

I don't have that citation. But Google doesn't spend billions on growing and maintaining apps just for public benefit. Those services pull their weight (now or prospectively) or Google would pull the plug. Generally their strategy is to drive traffic to search, and ads.

It's more subtle than just showing a traditional ad. When you search on a business (i.e. restaurant, hotel etc) in a given area, businesses are often paying to show up prominently in the results and/or when you're zoomed out.

They don't need to. They can show more relevant ads in search or email if they know where you are on the maps.

They have pricing plans for business users: https://mapsplatform.google.com/pricing/

I suspect if anyone uses it too much per hour or something, they will block you awhile and ask you to sign up for a plan. This is what google translate does.

Here’s the best description I’ve heard of why that’s a bad idea. It basically goes (paraphrased):

> Q: What would happen if you could look at someone and know who they are and where they live or work?

> A: A lot of women would die.

Google maps has some safeguards like certain areas have no imagery but really the biggest thing is it’s not fine detailed enough to be a threat to anybody. In fact, US law prevents (IIRC) sub-meter pixels on commercial satellite imagery.

In the US, most women who become violent crime victims are victimized by someone they know. More men are victimized by "stranger danger." At least according to the FBI’s uniform crime report. I’m skeptical this proposal would change existing crime patterns much.

Of course we can. A few nukes in the stratosphere will do the trick.

On a more serious note: I agree. Everyone should have that ability, then the worth decreases rapidly.

Still some players could act on the info more aggressively than others, but it would level the playing field to some extent.

So the concept of in dubio pro reo is dying faster and faster and all we do is shrug and tell ourselves we can't do anything about it.

I'm sorry, but this is a horrible take. I don't want every Joe Schmoe to have this capability in their back pocket. I can't even trust vetted people who have been elected into power to use it responsibly. Why on earth would I trust the drunk creep at the local dive with it? This is like saying every household in America should have an M1 Abrams because their value would rapidly decrease and therefore they'll be less dangerous.

> So Clearview has a database of billions of faces scraped from the internet, and sells subscriptions to match photos on that data.

I fully expect a class action lawsuit for Illinois residents soon. They need explicit opt in for biometric collection, regardless of how they obtained their photos.

See In Re: Clearview AI, Inc., Consumer Privacy Litigation (1:21-cv-00135)

I mean... I think it can be re-built from the ground up if the funding were there, but the type of people who build surveillance tools using nebulous data collection methods tend not do make a public service out of it.

Clearview has recently been getting into trouble for their practices in Canada: https://globalnews.ca/news/8451440/clearview-ai-facial-recog...

Edit: What they are doing here is illegal and they've since dropped operations in Canada. However they still track Canadian citizens and sell their services to anyone who wants to buy it outside of Canada. They've basically told Canadian citizens and the government here to go pound sand. I'm not sure how this gets resolved without new international agreements over privacy?

Arrest their directors, employees and investors upon entry into Canada.

Same with europe if this violates GDPR (likely).

We’d (rightfully or wrongfully) treat a cocaine cartel “shareholder” the same way.

It already has been deemed illegal in the EU:

https://noyb.eu/en/clearview-ai-deemed-illegal-eu

Your response reminds me of a case where a Russian citizen who operated an online gambling site went on vacation to some Caribbean Island. Except he never made it because the plane he was on made an unscheduled stop in Florida. The FBI also happened to be waiting for the plane to land and he got arrested. Anyone know what I'm talking about? It would've taken place around mid-2000s.

This guy: https://en.m.wikipedia.org/wiki/David_Carruthers

Whenever you change planes in USA, you are formally entering USA and subject to regular USA law.

The bigger case more recently was Belarus forcing down a Ryanair plane carrying an activist while overflying Belarusian airspace to arrest him:

https://www.voanews.com/a/europe_lukashenko-defends-divertin...

There’s also a Case out there of a birth on a flight between USA and Qatar being granted Canadian citizenship because the flight was over Canadian soil at the time of delivery:

https://en.m.wikipedia.org/wiki/Birth_aboard_aircraft_and_sh...

https://www.ctvnews.ca/mobile/canada/air-born-will-baby-deli...

yes, this is one of the reasons i never take flights that transit thru the US. their border situation is atrocious, and the fact that I have to enter their country for a transit is simply kafka-esque. instead of being a civilised country when it comes to transit they’re basically the belarus of north america.

To be honest, any country that wants to arrest you would have no issues with grabbing you from the transit lounge/terminal.

It costs the US airlines and airports a lot of business because US requirements to enter are difficult while other countries just use their brain and help out their airlines by permitting transit-without-entry.

That's it! No wonder I couldn't find it, I had half the details wrong. Online gambling, arrested mid-trip and in 2006. But British not Russian, Texas not Florida and no unscheduled maintenance. Funny how the mind fills in the blanks 15 years after passively reading a news article.

The Belarus case broke at least one big international agreement about planes. Was that even remotely the case for that situation in the USA ?

> Arrest their directors, employees and investors upon entry into Canada.

or sanction them. eg. https://en.wikipedia.org/wiki/United_States_sanctions#Person...

The worst part about Clearview is there is no way to easily opt-out if you do not live in California or Illinois.

I wrote Clearview an email a long time ago saying I would like my data removed. They asked for my ID to verify who I was. Clearview responds saying thanks for sending your ID, but you aren’t a California resident so we don’t have to remove your data. I wrote follow up emails still asking if they could consider removing it. After my follow up email, they just stopped responding. It’s been months now and no word back.

I get that they have no legal obligation to, but they should provide people a way to opt-out in other states. It felt like a bait and switch to ask for my personal information then just ghost me. However, I guess nobody can expect Clearview to be a moral company.

Not to mention you gave them even more (highly sensitive) personal data about yourself.

Typically with these types of requests it isn’t abnormal to have to give over an ID. I don’t like the idea of doing it, but usually it’s the only way to do it. Even if you’re a California resident, you’re forced to upload your ID to remove data. It is kind of a no-win scenario. Either they keep your facial data or you give over highly sensitive data to get the facial data removed. I only did it because it seemed initially they were willing to remove the facial data (despite their very delayed response).

Rent a cheap place in California or in the EU.

Establishing second residence isn't that expensive.

Is clear view requiring people to live in California or the EU for California/EU rights to apply??

Those state’s rights should apply for anyone present in those states (and possibly airspace) unless they idiotically wrote their laws.

Yes, Clearview requires it. Technically, the laws can only cover residents of those areas so that is how Clearview is able to require residency.

I like this idea of residing in CA - for a weekend or whatever on a yearly trip. What about not having an ID?

And pay CA income taxes. And pay exit tax in case you want to leave.

You only need to pay CA income taxes on CA income. If it is your second/vacation home and you don't earn money there, your tax home is obviously elsewhere, while still being a legal CA resident.

I wrote them an email saying I would like my data removed (I’m a california resident) and they asked for more data to help them remove me (a photo or something). It felt so backwards. Like they need to enter me into the database to ensure they don’t enter me into the database? If complying with the law results in paradoxes like that, the business doesn’t seem like it should be legal.

I never sent them the photo.

Yeah same, it feels very backward to me. I mean I can get that they want an ID to make sure you aren’t filing a request on someone else’s behalf without them knowing. Still, it is a bit disturbing they need an ID. Supposedly they don’t do anything with the ID info, but you never know I guess. I sent the ID picture in hopes they’d remove the facial data, because they did for others. However, when they denied removing the data I instantly regretted sending the picture. A bit of a stupid move on my part to be honest.

I requested they remove any images they may have of me as well, and got the same response, even though I pointed out that quite a few of the images of me are self hosted, all rights reserved, and they don't have permission to do anything but view them. They asked me the same thing which is insane, then silence.

Like what, am I going to send EVERY headshot I have to them? I can't wait for them to get lawyer nuked.

These are funny from an outsider perspective.

- Hi I'm John Smith (...)

- Hi John, could you send us your photos?

- (...)

- EOT

What did your city and state’s regulators think?

Just because clearview says fuck off because you’re not in California, your local people may not appreciate the fuck off so much.

Also, ask for what clear view has on you, not just removal. Sometimes they’re still obliged to correct any incorrect records even if they don’t have to legally remove them.

And don’t email them. Send a letter with a stamp with a CD. That’s how my government responds when I ask them for records.

That’s what’s interesting. My state has a law going into effect soon that covers requesting the removal of facial data, but at the moment it hasn’t started yet. Presumably once it goes into effect, Clearview would be legally obligated to remove my data.

Also, forgive me for asking this dumb question, but what do you mean by send a letter with a CD? What does CD stand for? Is it some kind of certificate of receipt? I tried looking it up, but didn’t find much about it.

Also, in terms of asking what they have on me, I have basically been unable to have any communication with them. Others seem to have the same issue. In general, Clearview takes months to respond to any sort of email (at least for me). I sent one semi-recently from a different email address. Hopefully they respond to that.

I assume they mean certified delivery: https://faq.usps.com/s/article/What-is-Certified-Mail

Nope, compact disc.

It’s really annoying since none of my computers have one and even a PlayStation can’t open pdfs.

But these are recipient problems.

And yeah, they even do it for 4mb files. Sometimes even with a custom sticker label. And they send me emails to say they’ve sent the records in the mail.

Though recorded/certified delivery is a good idea too.

That’s hilarious, that ID is part of your file now.

In Illinois, one has to opt in for biometric collection, not opt out. If they have anything on me, they're already breaking the law.

> The worst part about Clearview is there is no way to easily opt-out if you do not live in California or Illinois.

That obviously has a scammy vibe. If only Californians can opt out, then why is their database and their business not limited to California?

The fact that Clearview have such a comprehensive database in the first case is rather unnerving. Shouldn't there have been some sort of legislation to stop them? One private company having access to this information is not a good idea.

There's nothing really stopping anyone making such a database considering as I understand they just scraped faces from the public internet. There's no proof they were colluding with anyone.

For example, you could scrape and download every single public tiktok video and use that data as you see fit, if you can access it via a web browser it's fair game these days.

So copyright isn't a thing anymore?

I think the process of recovering damages should be standardized, like it is for music recordings: A fixed, but fairly stiff, penalty per offense. This would encourage bounty hunters to go after offenders, and split the proceeds with the owners. The act of offering to sell or share personal information would be interpreted as "intent," and automatically triple the damages.

Probably can’t sue for damages if you can’t prove damages, and probably can’t sue for statutory damages if you can’t prove they’ve got your IP (since it’s their database is an opaque box to you and me).

I’m not sure that’s true in the EU, for instance. It seems like it would fall foul of the GDPR, doesn’t it?

Clearview is compliant with GDPR FYI

That's an interesting position that is at least in dispute [1][2]. My prior based on how companies behave is that they are indeed in violation of the GDPR.

[1] https://techcrunch.com/2021/12/16/clearview-gdpr-breaches-fr...

[2] https://fortune.com/2021/05/27/europe-clearview-ai-gdpr-comp...

"legislation" only stops people acting in good faith.

Well, you gave it to them voluntary, by placing your photos on the public internet. "Dumb fuck", as Zuck put it.

This technology is so widespread, Apple could do it, Microsoft could do it, Google could do it, Facebook could do it, but for some reason we are focusing on Clearview. I think it's because only Clearview is selling it to law enforcement, thereby turning a previously expensive human task of identification and tracking, into a cheaper automated one. I'm reminded of the line 'With great power comes great responsibility'. The more people who have the power, the greater the odds that the responsibility will be abused in the pursuit of profit or even darker motives. History is littered with shithole autocracies that ran their corners of the world into the ground by eliminating economic participation and mobility using new advancements that delivered new concentrations of power. Here this technology gives smaller and smaller jurisdictions the power to turn themselves into absolute authorities, and it needs to be balanced with better oversight, better protections for people to move around without harassment. I hope this country is able to create and uphold new Federal legislation to that effect. Otherwise it's going to back an awful lot of people into a corner. I guess this is the big political question being fought out between the two main parties right now...

On Clearview's website they state they have 3 billion facial images scraped from the web. That seems to be a low cost of entry to accomplish the same level of service. I am a bit surprised that the FBI wouldn't have a similar database.

We are concerned about the other companies doing it. At least I am.

Of course the FBI would be collaborating on surveillance with the far-right, including weev & friends.

https://www.huffingtonpost.ca/entry/clearview-ai-facial-reco...

https://www.businessinsider.com/clearview-ai-far-right-white...

The FBI, CIA, NSA and every other alphabet agency down to the state, municiple level already have access to your face, address, d.o.b, phone info, income, bills, investments, physical and online activity.

Having the clearview info will help them build a better picture of who you hang with (all those photo's on FB, Snapchat, Insta, tiktok, etc) and the circles you keep. When they crunch all that data they'll be able to build a complete picture and remove any minutia of privacy you still might have had.

Here's an idea for you folks who are chasing the next big thing, develop something that will erase a person from the matrix.

Isn't this the acceptable nature of participating in a society? Covid was an opportunity to force people online - creating new accounts, accepting new privacy policies, getting used to Zoom calls and having more of your data legally recorded and used in ways you may not have considered (but you pressed "I agree" on that privacy policy!).

You can live off grid, but even that is very difficult with satellites overhead and trying to avoid every CCTV camera or digital device that may vacuum you up is bordering on impossible.

Unless you erase yourself from life, but even then someone will likely make a note of that somewhere ;).

Can't be too hard to make mismatching profiles with your face and different names and vice versa.

Clearview AI is disturbing. I have had a public conversation with one of the founders (who had left the company), where he said, "you civil libertarian types are all going to lose. We are going to steamroll you." And dismissed any and all concerns related to the technology.

They've taken the data without consent and sold it to police departments across the world with provable records of human rights violation, whilst claiming a 100% facial match accuracy. I am not exaggerating. They claim — to the police departments they sell to — that their technology is 100% accurate and doesn't have false positives. https://www.buzzfeednews.com/article/carolinehaskins1/clearv...

They are a combustible mix of greed and ambition with lies sprinkled into the mix. It astounded me that he had the gall to say and imply in a public setting that they are "civil libertarian types" (like me) were going to go the way of the dodo and that the totalitarian surveillance state was the way of the future.

I hope he is proven wrong. And I hope that they get sued and sanctioned out of existence like the NSO Group.

> public conversation Is there a verifiable record somewhere? I think there's no outrage because this is unknown to laypeople. However, I've very confident this would be a unifying issue for far left and right internet-outrage machines.

> "you civil libertarian types are all going to lose. We are going to steamroll you." And dismissed any and all concerns related to the technology.

That's pretty spot on, matches with my experience as well as every historical record ever written. We have a country right now that has installed labor/concentration camps and the rest of the world is dancing to their fiddle. Steamroll is an understatement. I don't want to know what is going on at those camps, but it wouldn't surprise me if many wish they could be steamrolled to end it all.

"Civil Libertarian Types" are abound on HN. They forget human nature and think that change is right around the corner. Sometimes I wonder if they're playing the devil's advocate or if they're really that stupid. It's bread and circuses of a very amusing and gullible sort.

These sort of people have no incentive whatsoever to change what they're doing. So yes, they're laughing at you as they fuck you over and profit from it. They know you're powerless, and they know you know it too.

What are you going to do about it? I'd genuinely like to see more than keyboard warring from you CLTs, but you're pretty useless about the whole thing.

> I hope he is proven wrong.

Was it the failed model that ran some phishing schemes and hired people to commit mass copyright infringement to build their database or the right wing politician that built ideological porn filtering software that blocked the ACLU and EFF?

Are those really the people we want controlling technology that's ethically questionable?

https://en.wikipedia.org/wiki/Clearview_AI

> I hope he is proven wrong.

Covid provided a massive boost for all of these kinds of systems. I say the cat is out of the bag. Just like when passports came to being “a thing” back in WW1. Before WW1 the British Empire convinced Belgium to drop the idea of passports. During WW1 governments across the world realised that using passports to control who can come to their country was easily sold to their citizens. 100 years later they are standardised across the planet. I can see the current trend of vaccine passports and the tracking in the same vein.

I've tried Clearview and it's more sizzle than steak. The images available were like a crippled duckduckgo search. There were photos available you didn't expect or find elsewhere. Primarily headshots. Shots with enough clarity are required which removes most poor quality facebook / myspace / instagram photos as the facial features are not clear enough and the resolution is poor.

It doesn't connect someone to a single identity and gather different social aliases and build a social graph.

It's another tool that simplifies a basic search. Honestly it's disappointing. I'd be worried about what acxiom is building. They can identity you anywhere online and connect you to other sites you visit.

The fact that the product sucks doesn't invalidate the badness of it existing.

This is an issue, but I'm curious, would it be possible to do a sort of 'denial of service' attack on the AI by overloading the internet with deepfakes? What if there are 50 deepfaked 'similars' to me out there?

Create dirty data and the whole dataset becomes less valuable. This would need some automation for dirty names and dirty faces and upload to public websites.

I don’t know what the statistics are for stolen license plates but it seems to me, here in the San Francisco Bay Area anyway, that whenever there’s a recorded incident of a robbery or assault where the perpetrators used a getaway car, in all likelihood they put stolen plates on the car.

Here’s an example of a gang who prowled around with impunity, comfortable that their cars could not be traced because they would just put stolen plates.

https://sfist.com/2021/12/15/six-bay-area-men-arrested-for-a...

Technology like Clearview could find a use case not just for facial recognition but vehicle recognition with validation to the license plate. We have many checkpoints and toll junctions where it can be installed and just run these exception checks. At the very least it would establish statistics on these gangs that are committing crimes with impunity. Vehicle recognition would be much easier than facial recognition.

I think in 10-20 years they’ll have this and more. Imagine dash cam videos from cop cars being analyzed and correlated by software so that someone who sped in between two cop cars 100 miles apart gets ticketed because they could not have got from one to the other without speeding.

The match between what Clearview AI is doing and it's founder's personal stories is just... cinematic.

https://en.wikipedia.org/wiki/Hoan_Ton-That

What's it say about the rule of law in the USA when the leading domestic federal organization for investigation of criminal activity is fine using databases obtained illegally?

What's it say about the police being subordinate to the law?

Can you source the claims of databases being obtained illegally? An interesting conversation of "consent" comes to mind - I think the issue is if somebody's face appears in a news article is it allowed to then download that same image and store it for unrelated purposes? Wouldn't that make google image search illegal? How about FB's now defunt face matching system they just removed? As far as I have seen some regulators have demanded the data be deleted as it might be considered to breach GDPR, but there are absolutely countries (like the USA) where this is considered absolutely acceptable behavior.

So far, no public court cases with decisions can be found anywhere from what I can see.

Clearview is selling use of the imagery without a license. If you or I scraped everything off of Instagram and started offering derived data as a SaaS, we'd go to jail.

These people don't because they're using it to help the cops.

They are selling the images themselves... if you scraped all the images off Instagram to make a list of the most popular memes and songs of the week the lawsuit would be much more difficult against you.

I don't think that is true. Scraping Facebook and the like has already been litigated several times. I can't find the outcome of those cases, but jail is definitely not a possibility.

If I take a picture of my family and tag everyone on Facebook, I own the copyright to that picture and the metadata. If Clearview AI scrapes that image and uses that to build their database, they're violating copyright.

Why don't they get prosecuted for industrial scale copyright infringement? IMHO it's because they're not violating the copyright of an ultra wealthy company or individual. It's only normal people getting cheating, so nothing gets done.

Copyright violation is a civil matter, not a criminal one. You don’t have D.A.s prosecute people for civil crimes, you sue them.

Copyright violation at scale across state lines is indeed a criminal matter, as the MPAA likes to remind you before every showing of a film.

You gave facebook permission to give the photo away. Facebook makes it available to the public.

Google, Yandex, Bing will read/save the image and provide a link to the original resource during a search if any text in a link matches. Uploading a photo will these searches will try to match against yours.

You as the copyright holder had the freedom to do whatever you wanted. By making it public you granted people access to view your content. People can legally view that image and make judgements including that you are similiar to another photo that was taken.

It's not about authors wealth your friends can save those images. If they sold or republished you could have a case but not if they directly link to your facebook pictures.

Yes, but Facebook, Google, Microsoft, Clearview or anyone else cannot take an image of me and use it for biometric purposes without my express consent as an Illinois resident. It doesn't matter if I took the photo or uploaded it. It's still a photo of me, and I have not consented.

> By making it public you granted people access to view your content. People can legally view that image and make judgements including that you are similiar to another photo that was taken.

This is not how copyright works. Clearview has no license to the copyrighted material from which they are profiting.

Facebook et al do - the license grant is in the signup TOS.

It might not actually be a violation of copyright, or search engines, the internet archive (which is nowhere near rich or powerful BTW), caching services and more in general could not legally exist in a practical manner. I'm guessing clear view exists under similar law. It's an image search engine, focused on faces. Google & microsoft could probably implement it today with their own image index, but don't because they will get a shit ton of negative PR and legislation for not enough money for that niche use case, because they are too big and powerful.

Google could very easily start the most effective hedge fund in the world due to their search engine power and ability to hire the smart, but do not, probably for much of the same reasons.

This is a great way to get a paranoid schizophrenic to send mail bombs to your company. Anyone who works at this company is utterly morally bankrupt and anyone with their morals intact should blacklist them.

https://sandlab.cs.uchicago.edu/fawkes

Facial recognition software.

I'm sure the FBI would go into hysterics if there was a publicly accessible facial recognition database of all FBI employees and all FBI informants - but is there really a problem with that? Why should secretive police organiziations - which have a notorious history related to the establishment of totalitarian states - be subject to different rules about data privacy then the general public is? Why shouldn't citizens have access to an app that allows them to easily identify undercover police seeking to infiltrate peaceful protest movements via facial recognition, for example?

The counterargument (authoritarian) is that police spies are necessary to combat various forms of crime in which all parties to the criminal activity have no interest in reporting said criminal activity - for example, illegal drug transactions. The counter-counterargument (libertarian) is that such forms of crime are actually victimless and the solution is to legalize and regulate all such drug transactions, to move them into the same sphere as alcohol sales.

It would be a very curious world, however, in which anyone could point their phone at anyone else and get an immediate background report on them (which is apparently what the government agencies like the FBI want to be able to do), including their home address, phone number, credit history, educational background, criminal records if any, online browsing habits, family and business relationships, medical history, travel patterns... that would be a very creepy world indeed. It already seems to exist for the secret police, however, with tools like Palantir coupled to this facial recognition system. The potential for gross abuse is pretty obvious.

Government regulation banning all such facial recognition software and data collection and aggregation is the only realistic option if we want to prevent such a dystopian future. People should have a right to privacy, even if this destroys the business model of Google, Facebook, and other personal data-based targeted advertising platforms, and so what if it makes the job of law enforcement harder? Read the Bill of Rights, about half of it is all about making it harder for law enforcement to snoop on citizens.

Information ownership and a digital bill of rights would solve this. It should be considered a basic human right to own one’s information just as one owns one’s body.

Yup, police always want the power relationship to be one way. You see it today with special public records exceptions, special license plates and more, which are not available for many others who would benefit, like DV victims, psychologists, and more.

Reply on news.ycombinator.com