ah, sarcasm. when you're desperate and have nothing else.
Last edited by a moderator:
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Remains Silent About Plans to Detect Known CSAM Stored in iCloud Photos
- Thread starter MacRumors
- Start date
- Sort by reaction score
ah, sarcasm. when you're desperate and have nothing else.
Apple is going to do whatever it thinks will maximize shareholder wealth regardless of the ethical issues involved.
And there we have perfectly summed up everything that’s wrong with capitalism and particularly surveillance capitalism.
I agree that implementing this functionality is a bad idea, but you’re misinformed on how this implementation works. This functionality doesn’t use computer vision to, for example, detect children in a state of undress. Rather, it creates a derivative of each photo uploaded to iCloud and compares it to similarly created derivatives of known CSAM images by checking their hashes.
A photo taken by a parent of their kid in the bathtub or whatever will almost definitely not match a photo in the database; even in the vanishingly small chance that there is a false match, you would have to meet a threshold (reportedly around 30) of matched images and a human would review the matches before forwarding the matter to authorities.
The implementation is fine server-side; it works and has put a lot of bad people in jail while allowing people to simply not use a cloud service if they don’t want their photos scanned. Building it into the OS is another matter — how likely is Apple to resist government pressure to add custom databases including non-CSAM content or reapply it for features other than iCloud Photos? I think we know the answer.
So, you’re saying that, knowing how the US government CURRENTLY uses tech products to spy on people, they are currently NOT using tech products to spy on people where it comes to unencrypted images stored in cloud image repositories. So, they’re currently using tech products to spy, but, where it makes the conspiracy convenient, they’re absolutely not in one very specific area?
The government currently (supposedly) doesn’t require Apple to monitor for the US government because the government is currently using tech products to spy on people. BUT, the government, with its tech spying, has just been WAITING to hand off the tech spying to Apple and, I suppose, Google, Samsung, Microsoft, etc.? Or, is it that the government ONLY wants to hand off the tech spying to Apple in this case and most certainly not anyone else?
I’m no fan of Antonin Scalia, but he was exactly correct when he said that, with respect to U.S. law, the criminality of a few is sometimes insulated in order to protect the privacy of us all. The people trafficking in this stuff need to be locked up, but that doesn’t mean that every measure taken against them is acceptable in the face of real risks for collateral damage to everyone’s privacy.
The issue here is that they're using AI to generate the hashes, it isn't simply a scan for identical hashes of photos. You don't need any AI to compare hashes, so what they're doing and how effective it is all speculation.
Bottom line is this system would use AI on your phone to scan your content, somehow. Given how primitive AI seems to be recognizing objects when using the camera, my speculation would be that it would make lots of mistakes.
And how long before this same technology is used to identify terrorism, and how effective and prone to mistake/bias would that be? Imagine what they would train the AI with to find terrorists--brown skinned, young men?
Just not a good idea to have our tech reporting on us to the government. The government has access to amazing technology itself, let them use their own resources and not use our computers against us.
Bottom line is this system would use AI on your phone to scan your content, somehow. Given how primitive AI seems to be recognizing objects when using the camera, my speculation would be that it would make lots of mistakes.
And how long before this same technology is used to identify terrorism, and how effective and prone to mistake/bias would that be? Imagine what they would train the AI with to find terrorists--brown skinned, young men?
Just not a good idea to have our tech reporting on us to the government. The government has access to amazing technology itself, let them use their own resources and not use our computers against us.
They're using AI to generate the hashes, not simply hashing the image file and doing a look up. The old style of hash the image and compare to known hashes would require no AI. I don't know if that's a great idea, but this implementation is more concerning because we don't know how it works at all--it's not just reporting for identical photos.
Traditional method was: Image>hash function>hash>compare to known bad hashes
Apple method: Image>AI scan>AI output>hash function>hash>compare to known bad hashes of AI output
Last edited:
People are judging this based on its current purported modus operandi but forgetting that everything changes, nothing stays the same. Today it’s child abuse. Tomorrow it’s anti public policy memes. And anyone who thinks this can’t happen has probably been in a coma for the past two years.
Actually, I can tell you right now, no future Snowden required.
The government will be using newer, even more advanced technology to spy on people. And will, very likely, still NOT be handing off their surveillance to any tech companies.Let parents enable it to protect their children’s devices.
That’s the solution.
If others upload abusive images to iCloud, perform the hash detection automatically on the server and report them to the FBI.
Done.
That’s the solution.
If others upload abusive images to iCloud, perform the hash detection automatically on the server and report them to the FBI.
Done.
They're going to be scanning everything. Not just Apple obviously. The future is everything is analyzed a million times over...and it will be very helpful in many cases. Like you want a picture of your friend Jimmy? Tag him once and the AI can find all old pics of him, etc. But again, don't think it's a great idea to have these scanning features reporting back to the government.
So, your idea is that, in these countries, the governments ABSOLUTELY do not have the keys to decryption OR the ability to request any images they want to see, right now? All image repositories are able to be decrypted by the service providers so they can perform CSAM scans today, so they’ll decrypt for their own purpose but NOT decrypt for the government if the government asks, even though they’ll give the government the matched CSAM hash matches when they ask?
No, this is about creating a technology that can be used to perform all the worst case scenarios from 1984. I think we can begrudgingly agree that Apple has the right to scan images on their own service - iCloud. But we will NEVER agree upon Apple creating a process to scan people’s personal devices and report them to the authorities.
Are you seriously saying that you can’t see anything wrong with doing this?
Famous last words there, in bold.
I think this is correct, too. Like we have to have some privacy and control over our own information. I think what's on our own hard drives/ssd, we should control. Once we upload it to Apple's servers, then I think we lose that control.
And soon enough everything will be networked storage probably, so we may have to revisit the idea of whether it makes sense on Apples servers too, but that's for a later date.
Ironic that the 1st example everyone uses for obvious abuse, is your beloved communist China - surveillance communism to the 1000th degree.
if you get to claim that REAL communism has never been tried, I’d contend that we don’t have capitalism, we have corporatism(corporation based, not classical definition) with the government bought and paid for by corporations, instead of REAL capitalism.
From where do you get the idea that I love China? China is also as capitalist as they come, with a totalitarian regime. China is a great example of what the U.S will become things going the way they seem after the next election.
And you’re right that we don’t have have real capitalism in the west. That was tried in the third-world. And that’s why it is now called the third-world.
This is key. The method used for hashing CSAM literally adds nothing to the volumes of information any government already has on anyone through far more effective methods that we don’t even know about.
No. A visual derivative is created of the image before hashing in order to thwart trivial workarounds to prevent a hash from matching (making a color photo grayscale, cropping, etc.). AI isn’t needed for that step and isn’t used for that. Computer vision is far too imperfect to be used for a case like this, where too many false matches can ruin someone’s life.
Apple did implement a computer vision-based feature along these lines, but it’s strictly on-device, unrelated to iCloud Photos, and only for children in Family Sharing. It warns them about sensitive photos in iMessage.
Some overzealous clerk saw baby bath photos and jumped the gun? The entire lot should have been taken to court.
Not remotely the same.
The overzealous clerk is now embedded in your phone. And, realistically, it's as dumb as someone working in a photo processing lab.
So, your thinking is that cloud providers should not scan for such images because no one would ever use cloud providers for this purpose, yes? What’s more likely is that they don’t use the services because they know they’re being scanned.