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

Hash collision in Apple NeuralHash model

github.com

Post reply on HN
Print view View original

401–410 of 725 posts

Re: Hash collision in Apple NeuralHash model

#401

Post by mrits »

newArray wrote: post #395

Hash collisions happen by design in the perceptual hash, its supposed to give equal hashes for small changes after all. Something I find interesting is the necessary consequences of the property of small edits resulting in the same hash. We can show that this is impossible to absolutely achieve, or in other words there must exist an image such that changing a single pixel will change the hash. Proof: Start with 2 ima…

The best part of this is we could all be flagged in a database for things like this and not even realize it. Who knows what kind of review process the governments are actually keeping around these

Re: Hash collision in Apple NeuralHash model

#402

Post by sandworm101 »

snowwrestler wrote: post #389
sandworm101 wrote: post #375

Earlier quoted context omitted.

>> They can’t use a warrant to force you to videotape someone Maybe in the narrow context of US domestic child pornography investigations. In the wider world it is very possible for police to force such things. Even in the US, CALEA demands that certain companies develop abilities that they would not normally want (interception). The principal that US companies need not actively participate in police investigations d…

Can you explain why you think CALEA would apply to what Apple announced? And why such application would have had to wait until Apple announced this? In other words, if the law can force Apple to do things in general, why does the law need to wait for Apple to announce certain capabilities first?

>> why you think CALEA would apply to what Apple announced?

Read what I stated. I said no such thing. I said that CALEA shows that companies can sometimes be forced to be do things they do not want to do, to take actions at the behest of law enforcement that they would not normally do. CALEA would clearly not be applied to apple in this case. It would be some other law/warrant/NSL that would force apple to do something it normally would not do. CALEA stands as proof that such things have long been acceptable under US law.

Re: Hash collision in Apple NeuralHash model

#403

Post by ballenf »

hughrr wrote: post #7

That’s end game. Now you can use it for targeted attacks against innocent people. This needs to be shut down and disposed of immediately. There is no other outcome which is socially acceptable for Apple. I feel vindicated now. There are a lot of people saying that I’m insane as I’ve dumped the entire iOS ecosystem in the last week. But Craig was busy steamrolling out the marketing still only a couple of days back abo…

Is it theoretically possible that photo filter effect could raise the false positive rate to make the 1 / a trillion number off by several order of magnitude?

Could be either malicious or accidental (less likely presumably).

Re: Hash collision in Apple NeuralHash model

#404

Post by short_sells_poo »

stetrain wrote: post #341
short_sells_poo wrote: post #334

Earlier quoted context omitted.

Yes of course it is more private than outright scanning all photos but how does that relate to the ground state of private photos simply not being scanned at all? And please do not bring in whataboutist arguments like "but other cloud providers are already doing it". I'm honestly taken aback by how many people on HN are completely OK with what Apple is pulling here. In my mind, it's irrelevant that the current system…

I don't think "All user data on device and in the cloud is not scannable for CSAM or retrievable with a warrant" is a tenable position in the current US political landscape, and even less so in other countries. And my point is that if the government wanted a dragnet they could just legislate or secretly order one. Just like they have done in various forms over the last 20 years. And Apple might not even be allowed to…

> "All user data on device and in the cloud is not scannable for CSAM or retrievable with a warrant" Who said anything about warrants? As far as I know the proposed system is proactive and requires no warrants at all.

More to the point, anyone remotely sophisticated can just encrypt CSAM into a binary blob and plaster it all over the cloud providers servers.

Ie, this system will possibly catch some small time perverts at the cost of even more potential of government misuse of the current technical landscape.

I consider it a similar situation to governments trying to legislate backdoors into encryption. A remotely sophisticated baddie will just ignore the laws, and all it does is add risks for innocents.

> And my point is that if the government wanted a dragnet they could just legislate or secretly order one. Just like they have done in various forms over the last 20 years. And Apple might not even be allowed to tell us it is happening.

I don't understand how is this an argument against the pushback at all. So just because it could be worse, we should just throw our hands in the air and say "Oh it's all fine, it could be worse so the (so far) mild intrusion into privacy is nothing to worry about."

The thing is, these things seem to happen step by step so the outrage is minimized. Insert your favorite "frog being boiled slowly" anecdote here. You don't push back on the mild stuff, and before you realize things have gotten so much worse.

Re: Hash collision in Apple NeuralHash model

#405

Post by floatingatoll »

ksec wrote: post #362
hughrr wrote: post #7

That’s end game. Now you can use it for targeted attacks against innocent people. This needs to be shut down and disposed of immediately. There is no other outcome which is socially acceptable for Apple. I feel vindicated now. There are a lot of people saying that I’m insane as I’ve dumped the entire iOS ecosystem in the last week. But Craig was busy steamrolling out the marketing still only a couple of days back abo…

I actually want Apple to stand ground and implement this feature. Like you said the double down on PR and marketing was enough for me. I may not be dumping all iOS and Mac for now. But it was " the " definite signal and evidence this is no longer the old Steve Jobs's Apple. It is like watching Mark Zuckerberg talking about privacy when he doesn't understand anything about it. ( Or more like he has a different underst…

Steve Jobs would have implemented this in secret and never told us at all, same as they already did for iCloud photos so many years ago. That would have been a far better approach than today’s Apple is taking. Oh well.

Re: Hash collision in Apple NeuralHash model

#406

Post by Shorel »

hughrr wrote: post #7

That’s end game. Now you can use it for targeted attacks against innocent people. This needs to be shut down and disposed of immediately. There is no other outcome which is socially acceptable for Apple. I feel vindicated now. There are a lot of people saying that I’m insane as I’ve dumped the entire iOS ecosystem in the last week. But Craig was busy steamrolling out the marketing still only a couple of days back abo…

That's just a checkbox that indicates further action is required. All hashing algorithms have potential collisions.

Now, the hashes' database is not public, AFAIK it is in the hands of the FBI.

Re: Hash collision in Apple NeuralHash model

#407

Post by raxxorrax »

kmbfjr wrote: post #76
zaptheimpaler wrote: post #32

This is so overblown. Scanning images for CSAM seems to be a requirement followed by Facebook, Google, Insta and Snap already [1]: > To put this in perspective, in 2019 Facebook reported 65 million instances of CSAM on its platform, according to The New York Times. Google reported 3.5 million photos and videos, while Twitter and Snap reported “more than 100,000,” Apple, on the other hand, reported 3,000 photos. ALL o…

It is not overblown. I refuse to be a perpetual suspect in the possession and transfer of child pornography. Having these checks on my phone is just a short hop to checking everything the camera sees. I worked with a guy who was convicted of this very crime. Do you know what the FBI installs on his computer and phone? This kind of monitoring. I am not going to be treated like I am on parole.

To form a collision here you basically use a partial image of the original. What it can do is leave some conclusion about the workings of their blackbox algorithm as the collision image ripped those parts that probably have no influence on the resulting hash. There will likely be more though. As long as you don't have the original images, it will be hard to create "fake" ones.

Re: Hash collision in Apple NeuralHash model

#408

Post by short_sells_poo »

robertoandred wrote: post #331
short_sells_poo wrote: post #311

Earlier quoted context omitted.

Yes I'm aware that it checks against a known database but clearly there can be collisions. So eventually it will share someone's private images.

Or rather, it will share some gray blob apparently.

What is the point of sharing a gray blob? How is that going to prove anything?

Re: Hash collision in Apple NeuralHash model

#409

Post by Griffinsauce »

_jal wrote: post #153
coldtea wrote: post #135

Earlier quoted context omitted.

Accusations must be proven true, not untrue by the accused. And besides the "victim" in the latter case there was a whole lot of diplomatic pressure and political commotion to set him up, with carrots and sticks and the aid of friendly satellite states.

They must be proven true beyond a reasonable doubt to get a person in a funny robe to put them in jail. Normal humans are not required to prove anything in order to think them.

*in the US

Re: Hash collision in Apple NeuralHash model

#410

Post by sandworm101 »

snowwrestler wrote: post #381
sandworm101 wrote: post #363

Earlier quoted context omitted.

They are scanning images on online accounts, stuff that is stored on their systems and system files on machines (files they own). They are not, from what I have read, scanning customer-owned images stored on customer-owned devices.

Every AV (antivirus) software system scans customer-owned files on customer-owned devices. So the question is the same: if it’s easy for law enforcement to deputize such a system, where is the flood of cases built off of evidence gathered this way?

>> Every AV (antivirus) software system

Mine doesn't. If Ubuntu or ClamAV is scanning all my photos and reporting the results to Canonical against my will then I will soon be having words with Mr. Linus.

Post reply on HN