Re: Apple accused of underreporting suspected CSAM on its platforms

On 2024-07-27 09:55, Jolly Roger wrote:

On 2024-07-26, Alan <nuh-uh@nope.com> wrote:

On 2024-07-26 15:14, Jolly Roger wrote:

On 2024-07-26, Alan <nuh-uh@nope.com> wrote:

On 2024-07-26 09:11, Jolly Roger wrote:

On 2024-07-26, Chris <ithinkiam@gmail.com> wrote:

On 24/07/2024 22:35, Jolly Roger wrote:

On 2024-07-24, Chris <ithinkiam@gmail.com> wrote:

Andrew <andrew@spam.net> wrote:

Chris wrote on Wed, 24 Jul 2024 07:20:19 -0000 (UTC) :
>

The NSPCC should really be complaining at how ineffectual the
tech companies are rather than complain at Apple for not sending
millions of photos to already overwhelmed authorities.

>
For all that is in the news stories, it could be ZERO convictions
resulted.
>
Think about that.
>
Is it worth everyone's loss of privacy for maybe zero gain in
child safety?

>
Apple's solution wouldn't have resulted in any additional loss of
privacy

>
Actually, Apple could not guarantee that, and there was a non-zero
chance that false positive matches would result in privacy
violations.

>
True. The balance of risk was proportionate, however. Much moreso
than the current system.

>
Absolutely. I'm just of the opinion if one innocent person is harmed,
that's one too many. Would you want to be that unlucky innocent
person who has to deal with charges, a potential criminal sexual
violation on your record, and all that comes with it? I certainly
wouldn't.

>
Except that Apple's system wouldn't automatically trigger charges.
>
An actual human would review the images in question...

>
And at that point, someone's privacy may be violated. Do you want a
stranger looking at photos of your sick child? What if that stranger
came to the conclusion that those photos are somehow classifiable as
sexual or abusive in some way? Would you want to have to argue your case
in court because of it?

>
Yes. At that point...
>
...if and only if the person is INNOCENT...
>
...someone's privacy is unnecessarily violated.
>
And it's a stretch to imagine that:
>
1. Innocent pictures would be matched with KNOWN CSAM images, AND;

 Not it's not. There was a margin of error in the proposed matching
algorithms.

I'm not saying it's impossible. Just very unlikely.

 

(the logical AND)
>
2. A person reviewing those images after they've been flagged wouldn't
notice they don't actually match; AND

 That decision is a human one, and humans make mistakes and have biased
beliefs that can lead them to make faulty decisions.

I'm not saying it's impossible. Just very unlikely.

 

3. The owner of those images at that point would be charged when they
could then show that they were in fact innocent images.

 Innocent people shouldn't have to prove anything to anyone.

We've already had two very statistically unlikely events that had to happen to get to this point.
At this point, it's pretty much equivalent to the search warrant. Information that suggests a crime might have been committed has been communicated to the authorities. And while someone's pictures have been examined by a human, that person doesn't know they've been examined, so where is the damage?

 

Yes, but one is one too many in my book.

>
And yet you are fine with innocent people's privacy being violated
when a search warrant is issued erroneously.

 Search warrants require probable cause and are signed by a judge.
Totally different scenario.

But innocent people do get searched...
...and you literally just said:
'Innocent people shouldn't have to prove anything to anyone.'