Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

github.com

Post reply on HN
Print view View original

181–190 of 363 posts

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#181

Post by Syonyk »

simondotau wrote: post #164
beachy wrote: post #112

Earlier quoted context omitted.

This where the "fog of war" kicks in. What with doors being busted down, police departments making press releases, etc. I can easily imagine that the victim could be prosecuted, convicted and sent away because no-one understood the subtlety that their legal porn was not in fact CSAM.

The fog of war is largely in the realm of post-puberty minors, photos of which are not being included in Apple's corpus of hashes. I find it difficult to believe that anyone could mistake or otherwise "fog of war" a photograph of an adult and a prepubescent minor. And that's assuming someone develops a hash collision which doesn't substantially mangle the photograph like the example offered on Github. Specifically, o…

> Specifically, only images categorised as "A1" are being included in the hash set on iOS.

Do we know that for sure?

Apple has changed their mind enough times in the last week and a half that I'm convinced they're in full on defensive "wing it and say whatever will get people off our backs!" mode.

You can't read the threat modeling PDF and conclude that it was run through the normal Apple document review process. It reads nothing like a standard Apple document - it reads like a bunch of sleep deprived people were told to whip it up and publish it.

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#182

Post by simondotau »

Alupis wrote: post #50
Spivak wrote: post #49

Earlier quoted context omitted.

The FBI off their back that they aren’t doing enough to stop the spread of CP.

I find it hard to believe CSAM was so pervasive on iDevices that they'd feel compelled to do something about it. As far as we know (and I'm sure lots of eyeballs are looking now) Android doesn't do this. And frankly, why would Apple care that the FBI isn't cozy with them. Their entire brand is "security and privacy", kind of goes against most 3 Letter Agencies anyway.

I agree CSAM isn't likely to be pervasive in the photo libraries on iOS devices.

Android does not do on-device scanning, but Google does scan photos after they are uploaded to their cloud photo service. It's not on-device scanning, but the effect is functionally identical: photos that are being uploaded to the cloud are being scanned for CSAM. The only real distinction is who owns the CPU which computes the hash.

I doubt it's the FBI pressuring Apple. My suspicion is it's fear of the US Congress passing worse, even more privacy-invading laws under the guise of combating CSAM. If Apple's lobbyists can show that iPhones are already searching for CSAM, arguments for such laws get weaker.

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#183

Post by Syonyk »

ec109685 wrote: post #175
vermilingua wrote: post #37

The integrity of this entire system now relies on the security of the CSAM hash database, which has just dramatically increased in value to potential attackers. All it would take now, is for one CSAM hash to be known to the public, then uploading collided iPhone wallpapers to wallpaper download sites. That many false positives will overload whatever administrative capacity there is to review reports in a matter of da…

No, there’s another private hash function that also has to match the known CSAM image for an image to be considered a match. That one can’t be figured out through this technique.

They couldn't possibly be using the "other perceptual hash algorithm commonly used for this stuff," PhotoDNA, could they?

I mean, hopefully not, but at this point, it's reasonable to call just about everything into question on the topic.

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#184

Post by Alupis »

simondotau wrote: post #182
Alupis wrote: post #50

Earlier quoted context omitted.

I find it hard to believe CSAM was so pervasive on iDevices that they'd feel compelled to do something about it. As far as we know (and I'm sure lots of eyeballs are looking now) Android doesn't do this. And frankly, why would Apple care that the FBI isn't cozy with them. Their entire brand is "security and privacy", kind of goes against most 3 Letter Agencies anyway.

I agree CSAM isn't likely to be pervasive in the photo libraries on iOS devices. Android does not do on-device scanning, but Google does scan photos after they are uploaded to their cloud photo service. It's not on-device scanning, but the effect is functionally identical: photos that are being uploaded to the cloud are being scanned for CSAM. The only real distinction is who owns the CPU which computes the hash. I d…

> Android doesn't do on-device scanning, but Google does scan photos after they are uploaded to their cloud photo service.

So did Apple, and pretty much all cloud hosting providers.

This, on device, scanning is what's new, and very out of character for Apple.

> If Apple's lobbyists can show that iPhones are already searching for CSAM, arguments for such laws get weaker.

I'm not aware of any big anti-CSAM push being made by Congress. CSAM just isn't really a big issue in the US, the existing laws, and culture, are pretty effective already.

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#185

Post by fay59 »

sigmar wrote: post #151
fay59 wrote: post #123

Earlier quoted context omitted.

Before they make it to human review, photos in decrypted vouchers have to pass the CSAM match against a second classifier that Apple keeps to itself. Presumably, if it doesn’t match the same asset, it won’t be passed along. This is explained towards the end of the threat model document that Apple posted to its website. https://www.apple.com/child-safety/pdf/Security_Threat_Model...

What happens if someone leaks or guesses the weights on that "secret" classifier? The whole system is so ridiculous even before considering the amount of shenanigans the FBI could pull by putting in non-CSAM hashes.

For better or worse, opaque server-side CSAM models are the norm in the cloud photo hosting world. I imagine that the consequences would be roughly the same as if Google's, Facebook's or Microsoft's "secret classifiers" were leaked.

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#186

Post by Syonyk »

heavyset_go wrote: post #130
fnord77 wrote: post #111

how long until they start scanning a device's framebuffer in realtime? why stop at CSAM? Pirated material like movies next?

> how long until they start scanning a device's framebuffer in realtime? Some smart TVs do automated content recognition so the manufacturers can spy on what you're watching and sell the data to the highest bidders.

Roku's "privacy" policy is a hoot to read for stuff like this.

It's basically, "If we've come up with a way to grab it, we do. And send it to our servers. And do what we want with it."

It literally includes:

> We may receive information about the browser and devices you use to access the Internet, including our services, such as device types and models, unique identifiers (including, for Roku Devices, the Advertising Identifier associated with that device), IP address, operating system type and version, browser type and language, Wi-Fi network name and connection data, and information about other devices connected to the same network.

Emphasis mine. They literally have given themselves permission to nmap your LAN and upload the results!

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#187

Post by jazzyjackson »

cookiengineer wrote: post #176
ec109685 wrote: post #172

Earlier quoted context omitted.

That’s not how Apple’s system works. It’s not an image classifier. Only actual images that are derivatives of known CSAM images (a database of 250k images) will match. Random images of kids will not match those at any greater frequency than any other image.

Counter-question: At what point is child porn actually child porn, socially and statistically speaking? If I share that picture of my child with my friends and loved ones on Facebook - at what "scale" is it considered to be added to that database as child porn? 1k shares? 10k? Who's the one eligible to decide that? The judicatives? I think this scenario is a constitutional crisis because there's no good solution to i…

I think you're underestimating the severity of child abuse by orders of magnitude. CSAM is a database of child rape, not child nudity.

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#188

Post by simondotau »

Alupis wrote: post #184
simondotau wrote: post #182

Earlier quoted context omitted.

I agree CSAM isn't likely to be pervasive in the photo libraries on iOS devices. Android does not do on-device scanning, but Google does scan photos after they are uploaded to their cloud photo service. It's not on-device scanning, but the effect is functionally identical: photos that are being uploaded to the cloud are being scanned for CSAM. The only real distinction is who owns the CPU which computes the hash. I d…

> Android doesn't do on-device scanning, but Google does scan photos after they are uploaded to their cloud photo service. So did Apple, and pretty much all cloud hosting providers. This, on device, scanning is what's new, and very out of character for Apple. > If Apple's lobbyists can show that iPhones are already searching for CSAM, arguments for such laws get weaker. I'm not aware of any big anti-CSAM push being m…

We do know that Apple has been scanning email attachments sent via iCloud email. I don't think it's ever been claimed that Apple has ever scanned anyone's iCloud Photo Library.

Ethics aside, on-device scanning has the benefit of Constitutional protection, at least in the USA. Because the searching is being performed on private property, any attempt by the Government to try to expand the scope of searches would be a clear-cut 4th Amendment violation.

(Whereas if the scanning is done in the cloud, Government can compel searches and that would fall under the "third party doctrine" which is an end-run around the 4th Amendment.)

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#189

Post by djrogers »

copperx wrote: post #68
dannyw wrote: post #30

Earlier quoted context omitted.

The "send known CSAM" attack has existed for a while but never made sense. However, this technology enables a new class of attacks: "send legal porn, collided to match CSAM perceptual hashes". With the previous status quo: 1. The attacker faces charges of possessing and distributing child pornography 2. The victim may be investigated and charged with child pornography if LEO is somehow alerted (which requires work, a…

It seems like I'm not going to sleep tonight. Sure, there is hyperbole in OP's comment (CSAM ransomware and automated law enforcement aren't a thing yet), but we're a few steps from that reality. Even worse, how long will it take until other cloud storage services such as Dropbox, Amazon S3, Google Drive et al implement the same features? Or worse, required by law to do so? This sounds like the start of an exodus fro…

> Even worse, how long will it take until other cloud storage services such as Dropbox, Amazon S3, Google Drive et al implement the same features? Or worse, required by law to do so

They already do this. Google and Facebook have even issued reports detailing their various success rates…

Re: Show HN: Neural-hash-collider – Find target hash collisions for NeuralHash

#190

Post by ec109685 »

Syonyk wrote: post #183
ec109685 wrote: post #175

Earlier quoted context omitted.

No, there’s another private hash function that also has to match the known CSAM image for an image to be considered a match. That one can’t be figured out through this technique.

They couldn't possibly be using the "other perceptual hash algorithm commonly used for this stuff," PhotoDNA, could they? I mean, hopefully not, but at this point, it's reasonable to call just about everything into question on the topic.

Their goal is to minimize false positives, so would not be in their best interest.
Post reply on HN