Hacker News Reader

Comment by ummonk | Hacker News Reader

This is great news for individuals who don’t want companies snooping on their private images and don’t want to get ensnared with legal issues due to false positives. It also ensures a barrier against governments demanding scanning for dissident material and the like.

But let’s not pretend it’s a move that doesn’t come with a cost. Scanning for CSAM by companies does catch a lot of child abusers and result in children being rescued from abuse, and it’s likely that Apple implementing such scanning would have done the same.

View on HN

Why stop at CSAM? Why not scan for all kinds of crimes?

IMHO, the answer is that the police work should be done by the police and not by consumer products.

For some strange reason the “think of the children” argument is alive and well and can be used for all kind of privacy busting.

Seriously, why we don’t think of the domestic abuse victims and mandate all time listing microphones then?

The scanning for known fingerprints in Apple’s scheme only happened sever side.

Unlike Google, Apple didn’t have access to all photos in an unencrypted form to do whatever they pleased with them. If and only if 5 photo fingerprints matched a known csam photo, could Apple view a thumbnail of the photo.

How could that possibly be thought of as worse than Google’s “we can see all your data all the time” approach?

The fingerprinting doesn't happen by magic, your device analyses your photos and sends the results to Apple for control and Apple alert the authorities once enough matches happen because once enough matches happen your photos are decrypted for human inspection.

Inconsequential implementation details. Your device still acts as a snitch for whatever the authorities are looking for. Today looking for CSAM match tomorrow can be environmentalism, abortion, meat products etc - depending on how the politics evolve.

Apple’s severs are the snitches and they can only snitch if you upload five matching fingerprints.

Fine to think that scanning for csam is wrong, but the fact that it’s based on fingerprints sent to the cloud doesn’t make their scheme worse than Google who has access to unencrypted photo data to do whatever they want with (as argued by many in this thread).

Trauma experienced in childhood results in individual’s psychology being messed up in various ways for the remaining life. This messed-uppedness is contagious: it spreads to descendants, peers (school bullying), subordinate people (teacher to student) and so on.

Since I believe that a lot of issues humanity has to deal with come from mental health issues (I like to make this point so I won’t belabor it), I also find “think of the children”, while often misused, on its actual substance not an argument to be ridiculed.

Today I am reminded about The Ones who Walk Away from Omelas by Ursula Le Guin a little.

I never said that it's not an important issue and people shouldn't bother for the wellbeing of children.

I was actually responding to your lamenting how this argument is alive and well. Do you notice how you are self-contradicting?

No one is arguing that child porn is not a heinous thing. The argument is that there are also many other crimes that are in the same ballpark of severity, so why only police this one?

My comment explained why I think it is in a different ballpark of badness, maybe you could respond to that.

I think the issue here is that CSAM is and often was used as an excuse for invasive privacy breaking measures. Nobody can "not" think of the children, so whenever someone would argue against a measure, all they can say back is "so you don't care about children then?"

And people start getting tired of this serious issue being used, not because the people who want these measures passed care, but because of whatever agenda those people are actually interested.

It is precisely because it is so severe, that it is difficult to say no to measures in favor of it. But that same fact, is abused by people with bad intentions.

I have no desire to promote privacy invasive measures, but like the comment that started the thread: we have to acknowledge the horrible epidemic of this abuse and that the measures proposed would have been effective against it. We should be mindful that detecting and, upon human review (can’t believe how much disinformation was spread on this front), reporting CSAM would have been a good thing if it was entirely side-effect free.

However, we (as a crowd) have decided that those side-effects would be even worse than CSA itself. And perhaps we were right, but it should weigh on our consciences as we were the utilitarianist arbiters that decided individual suffering was not significant enough, and we should not rest as if we’ve “won” because a big problem (to solve which Apple seemed even willing to put its reputation on the line) remains unsolved.

The slippery slope argument is no doubt a valid one. But Apple specifically implemented (almost) the CSAM scanning after being threatened by multiple politicians - notably Lindsey Graham - that if they didn’t find a solution that they would legislate a backdoor into the OS for law enforcement. I think Apple was doing everything they could to appease the demand without actually violating users’ privacy. While it’s clear they failed to please end users, the solution itself was pretty ingenious.

Edit: https://www.eff.org/deeplinks/2019/12/senate-judiciary-commi...

> Scanning for CSAM by companies does catch a lot of child abusers and result in children being rescued from abuse, and it’s likely that Apple implementing such scanning would have done the same.

A strong claim that still lacks any strong proof. Especially once you start defining the term "child abuser".

According to actual child protection organisations [1] encrypted communications play a very minor role in distribution and is barely relevant, and the biggest factor is not detection, but the lack of proper resources to actually investigate cases and leads. There is no shortage of hard leads to investigate.

It is also rarely acknowledged that are very large percentage of abusers are from the victims social circle and often family members [2]. Meanwhile the whole surveillance fetishists very much try to keep up the false image of children being kidnapped by strangers and such.

The whole idea that child abusers are incredibly hard to catch and are oh so well hidden is a fairy tale. It is also important to differentiate between 1st and second grade child abusers, those who actively abuse the child, and those who "only" encourage it by sharing/purchasing the results of the abuse (I'm sure there is a significant overlap, but I do not know).

So is there a cost? Sure. But let's not pretend that it is even the lowest hanging fruit, or that the people building these services and asking for them are even interested in actually improving the situation, otherwise they would throw their weight behind one of the dozens of other more effective ways to actually improve child protection instead of screaming for a perpetual noise machine that even further drains the already very scarce resources of child protection services.

[1]: https://netzpolitik.org/2022/massenueberwachung-das-sagen-ki...

[2]: https://en.wikipedia.org/wiki/Child_abuse

True; but as mainstream platforms keep fighting CSAM, the only "reliable" place to get it is specialized sites on the dark web, and governments have been quite competent at busting those and identifying their users. I much prefer them turning darkweb CSAM sites into honeypots, than them dragnetting huge portions of the population.

At least the harm is limited to those who deserve it.

This is true in general, but not in Apple's CSAM detections that have been abandoned. They were only scanning for images that matched a know hash of CSAM. This did nothing to prevent new CSAM from being created, only possession of existing material.

That still catches abusers, as abusers are often consumers of material.

Detecting known CSAM does help fight its creation.

There's a Venn diagram of people who possess of library of CSAM and are also active child abusers themselves. Detecting possession is one way to try to get people who fit into that overlap.

It can also help reveal criminal networks. Prosecuting possession of CSAM works like it does for other networked crimes like drugs. CSAM is material that is illegal and hard to get. People who possess quantities of it almost certainly acquired it from suppliers, who themselves acquired from suppliers, etc. Law enforcement's goal is to find and flip possessors to walk this network back to the source: people who are abusing children and documenting it to create new material.

They also hope to discourage its creation by suppressing demand and raising risks and costs for abusers who create it (in both cases, by the threat of prosecution).

> But let’s not pretend it’s a move that doesn’t come with a cost. Scanning for CSAM by companies does catch a lot of child abusers and result in children being rescued from abuse, and it’s likely that Apple implementing such scanning would have done the same.

Let's not pretend that scanning everyone's photos is the only (or even a good) way to catch child abusers. Child abusers have been convicted without CSAM scanning before, so I am sure it will continue to be possible without. Just scanning everyone's photos may or may not be a convenient way to catch many, but the social cost for that is too damn high.

There is no guarantee that fewer children would get abused by scanning for CSAM, since abusers would probably adapt and avoid scanned services, while the innocent majority of people would have lost their privacy.

I think it's good to acknowledge, yes, this is worth the cost (since even if it would have opened up massive potential for this system to be abused, massive privacy loss for law abiding citizens, etc., it's true it might have caught a few abusers). Just like how the physical world equivalent - Governments putting cameras in everybody's homes that are watched by AI and police can access at any time when something is flagged - would be far more open to abuse than they would be useful for stopping crime.

We must always remember that mass surveillance is never the most effective way to stop crime, and is rarely even that useful. You make the haystack far larger and finding the needles gets more difficult, and AIs, while impressive, generate huge numbers of false positives.

Freedom is rarely free of costs.

I’m a nudist. Scanning for CSAM has caused no end of grief to the community and even led families losing their kids. It’s police-state type crap and I’m glad to see it go.

Reply on news.ycombinator.com