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

The deceptive PR behind Apple’s “expanded protections for children”

piotr.is

Post reply on HN
Print view View original

571–580 of 595 posts

Re: The deceptive PR behind Apple’s “expanded protections for children”

#571

Post by heavyset_go »

nicce wrote: post #263
zionic wrote: post #77

Earlier quoted context omitted.

Yes it does, it uses fuzzy perceptual hashes not crypto hashes. So if your innocent baby pic looks similar enough to a previously tagged child abuse image then YES, it will flag you and send a copy to the feds. And before you correct me, the Apple employee will see a picture of your naked baby and hit “forward to NCMEC”, which… upon investigation is actually just the feds

Perceptual hashes are extremly accurate. You might need a twin and top of that somehow identical environment almost in pixel level. Are news filled with false-positive accusations by PhotoDNA, flagging wrong images in Google, Facebook, Instagram etc.?

> Perceptual hashes are extremly accurate.

No, they are not. Ask anyone who has worked in this space[1][2], including myself. They are incredibly common.

Two images that kind of look like one another will have similar or the same hashes. That is the point of perceptual hashing.

[1] https://news.ycombinator.com/item?id=28091750

[2] https://news.ycombinator.com/item?id=28110159

Re: The deceptive PR behind Apple’s “expanded protections for children”

#572

Post by tim333 »

farmerstan wrote: post #219

Whoever controls the hash list controls your phone from now on. Period. End of sentence. Apple has not disclosed who gets to add new hashes to the list of CSAM hashes or what the process is to add new hashes. Do different countries have different hash lists? Because if the FBI or CIA or CCCP or KSA wants to arrest you, all they need to do is inject the hash of one of your photos into the “list” and you will be flagge…

Couple of gripes there.

Apple basically controls you phone anyway and have done for years as they can issue patches and os updates.

Also you can turn iphotos off - I've never used the thing in spite of owning various apple devices. I do use Google photos and doubt they are much different in terms of checking for CSAM.

Re: The deceptive PR behind Apple’s “expanded protections for children”

#573

Post by zepto »

phkahler wrote: post #552
zepto wrote: post #115

Earlier quoted context omitted.

I agree with you, but I want to correct you - they are using an image analysis hash, not a cryptographic hash. However it doesn’t change the logic of your argument. They require multiple positive marches, and they also require a visual derivative of the CSAM images to match.

>> they are using an image analysis hash, not a cryptographic hash. However it doesn’t change the logic of your argument. Yes, I think it fundamentally changes it. A crypto hash would be looking for specific known CSAM images. Image analysis is error-prone and will lead to problems and false positives. I don't care how good it is, that's actually looking at the users images and making a judgement.

> A crypto hash would be looking for specific known CSAM images. Image analysis is error-prone and will lead to problems and false positives.

The hashes they are using are only looking for specific known images.

Frankly, you should look at the technical docs on the mechanism at this point. It’s not the kind of image analysis you think it is.

Re: The deceptive PR behind Apple’s “expanded protections for children”

#574

Post by __app_dev__ »

fortenforge wrote: post #137
querez wrote: post #2

I have a newborn at home, and like every other parent, we take thousands of pictures and videos of our newest family member. We took pictures of the very first baby-bath. So now I have pictures of a naked baby on my phone. Does that mean that pictures of my newborn baby will be uploaded to Apple for further analysis, potentially stored for indefinite time, shared with law enforcement?

Lots of people responding to this seem to not understand how perceptual hashing / PhotoDNA works. It's true that they're not cryptographic hashes, but the false positive rate is vanishingly small. Apple claims it's 1 in a trillion [1], but suppose that you don't believe them. Google and Facebook and Microsoft are all using PhotoDNA (or equivalent perceptual hashing schemes) right now. Have you heard of some massive i…

Even if that were the case this will now make it possible to "swat" or target someone by using spyware to send pics to someones phone. Obviously not many people have access to this type of spyware (think NGO Pegasus) but if law enforcement really wanted to target someone this way they will be able to do so.

For everyone else getting access to the phone and manually loading the pic would work.

Additionally it's possible to fool AI: https://slazebni.cs.illinois.edu/fall18/lec12_adversarial.pd...

Re: The deceptive PR behind Apple’s “expanded protections for children”

#575

Post by croh »

catears wrote: post #516
ekianjo wrote: post #444

> The worst part is: how do I put my money where my mouth is? Am I going back to using Linux on the desktop (2022 will be the year of Linux on the desktop, remember), debugging wifi drivers and tirelessly trying to make resume-from-suspend work? Oh come on. DOn't make it sound like it's that bad. Wifi is a solved problem for a long time now, and you can buy Lenovo, System76 or Tuxedo if you want to make sure 100% thi…

I recently spent a full day trying to debug wifi on my dual-boot desktop with "newish" hardware. The wifi would occasionally drop 100% of packets, for between 5 seconds and 10 minutes, and then go back to normal. I tried installing different drivers and following some askubuntu forum posts, but nothing seemed to work. I'm not an expert on hardware stuff so I haven't got the knowledge to dig deep and find what caused…

Make sure windows is shutdown completely, not in hibernate. It is not like wondows has "solved" wifi but it created problem in first place.

Re: The deceptive PR behind Apple’s “expanded protections for children”

#576

Post by citizenpaul »

This system has 0 transparency. I cannot appeal. I cannot check my status. I cannot check this mysterious "counter" I cannot even check an image i have to see if it is flagged. I cannot know who is manually looking at my data no matter how private it its.

I get a note in my fruit delivery with the name of the person and time they were at work packing my food. Yet I'm told that I cannot know anything about what is happening with at least a partially automated system that can potentially put me in jail for the next 20 years?

Re: The deceptive PR behind Apple’s “expanded protections for children”

#577

Post by phkahler »

daveidol wrote: post #287
shapefrog wrote: post #121

Earlier quoted context omitted.

You saying one conspiracy is 100% true but another one is laugh out loud impossible?

Wait, what conspiracy am I saying is 100% true? If your phone was listening to everything you do all the time then you'd need a lot of processing to analyze it on-device and exfiltrate the critical pieces of info (which would be pretty obvious when looking at CPU usage and battery life), or you'd be sending an insane amount of data off over the internet all the time. I'm not saying it's technically impossible - I'm j…

There is a possible explanation for the schizophrenia example. My wife did a google search for a particular medication that person had been on and went off of. That was a couple days prior to the ads showing up. So good old Google could have been part of that - not sure how google searches affect facebook ads - or perhaps it was the pages she visited to read about that medication. It's all just f-ing creepy sometimes regardless of how it come about. It sure seems like verbal conversations being overheard is a part of it.

Re: The deceptive PR behind Apple’s “expanded protections for children”

#578

Post by heavyset_go »

tpush wrote: post #563
heavyset_go wrote: post #562

Earlier quoted context omitted.

Not all transformations result in the same hash, hence why fuzzy matching is necessary. Some perceptual hashing techniques try to incorporate common transformations like rotations and scaling, but if you change an image slightly, but enough, hashes will be different but still similar.

Apple's system only concerns exact hash matches, and is built so transformed images hash to the same number.

> and is built so transformed images hash to the same number.

That is the goal of all perceptual hashing algorithms.

> Apple's system only concerns exact hash matches

Then it is almost useless. All someone would need to do to evade the system is make minor enough adjustments to illegal images so that the distance between resulting hash and the original is minor.

In reality, perceptual hashing systems use fuzzy matching on hashes by using a metric like hamming distance to calculate differences between two or more hashes.

Re: The deceptive PR behind Apple’s “expanded protections for children”

#579

Post by tpush »

heavyset_go wrote: post #578
tpush wrote: post #563

Earlier quoted context omitted.

Apple's system only concerns exact hash matches, and is built so transformed images hash to the same number.

> and is built so transformed images hash to the same number. That is the goal of all perceptual hashing algorithms. > Apple's system only concerns exact hash matches Then it is almost useless. All someone would need to do to evade the system is make minor enough adjustments to illegal images so that the distance between resulting hash and the original is minor. In reality, perceptual hashing systems use fuzzy matchi…

> Then it is almost useless.

I'm sure you're in a prime position to judge that.

> In reality, perceptual hashing systems use fuzzy matching on hashes by using a metric like hamming distance to calculate differences between two or more hashes.

Again, Apple's NeuralHash purposefully doesn't work like that [0].

[0] https://twitter.com/jonathanmayer/status/1423370142411476993

Re: The deceptive PR behind Apple’s “expanded protections for children”

#580

Post by mastazi »

dijit wrote: post #523
mastazi wrote: post #252

Earlier quoted context omitted.

While I don't expect any Linux phone to become "mainstream" any time soon, it would be good if we had at least one "polished" alternative available. PinePhone is still in beta and according to its own creators "aimed solely at early adopters"[1], while Librem 5 is experiencing supply chain issues with backorder shipping now scheduled to resume in October[2] There is a version of the Librem 5 which is made in USA and…

There was one but people wouldn't "put their money where their mouth is". The Jolla phone was made by ex-MeeGo/Maemo devs from Nokia, but nobody bought them. Now they're focusing on just the OS and they don't make any full fledged devices: https://jolla.com/

Yes, I knew about them, they're not the only open source phone that died out unfortunately. Hopefully things will change now that privacy has become something even the "average Joe" worries about.
Post reply on HN