You're missing my point. If Apple wanted to (or was forced to) install surveillance software on our devices, they could have done so at any point in time, and they wouldn't even be required to make some public announcement about it. You're right, iOS is closed source, so they could do this behind the radar.
Yet there are 33 pages of people complaining about this new tool, which is an infeasible target for abuse.
Yet I highly doubt the machine learning foundation and first few iterations of tools are built by tools themselves.
You're right in that the first iterations did not build themselves, which would imply that these tools could follow some Darwinian notion of evolution from primordial bits.
Not only that, but they claim they're only matching hashes from known pre-existing CP images, which means what they're actually doing is incentivizing the production of new images and the exploitation/trafficking of more children in order to produce images which aren't in the database and can't be caught until the next time the hashes are updated. They're making the problem of exploitation *worse* while introducing an attack vector to every iPhone. That's just stupid.
No, clearly Apple thinks they can use the outrage against this heinous crime to defuse any outrage over compromising the security of IOS. They've presented no evidence that this is a major problem or any real justification for implementing on-device scanning and compromising the security of the Messages app.
Yup. Machine learning into self evolving machines that can act and think like humans. But thats the topic for another day.
In fact, I assume they have done so already. It’s just that those tools have yet to find any reason to suggest prosecuting me for any “potential wrongdoing” at the moment.
Exactly. It incentivizes the creation of *new* child porn and the exploitation of more children in order to produce images whose hashes are not in the current database. It makes the problem *worse* and hurts *more* children. Why would anyone want to do that????
Apple's statements about the odds of erroneous flagging are meaningless without independent verification. Since the details of the technology are proprietary, that's impossible to do. They should open-source the scanning/flagging code so independent experts can validate it.
That "1 in a trillion" figure is just PR bafflegab unless it's independently verified.
It can be used for finding, for example, lots of near exact matches of movie frames, since that’s a straightforward addition to handling HEIF files. In other countries there are movies which are illegal to obtain that aren’t CP, plus there will inevitably be requests to use it as an anti-piracy measure.
I'll concede to these possibilities. Apple says they won't do anything aside from CSAM, but that's just what Apple says. In any case, while I wouldn't equate CP with pirated movies, both are still illegal to own, and the point of "don't do illegal stuff and you don't need to worry" still stands.
Apple doesn't need a warrant as it is not a a law enforcement agency.
I'm just curious what was preventing <insert evil actor here> from forcing Apple (potentially Apple itself is the evil actor) from doing this prior to this tool being implemented? iOS 14 includes powerful scanning techniques on all of your photos (and the rest of your phone) that far outperform this tool in terms of what can be found and how easy it would be to alter it. If <evil actor> is attempting to surveil, why not just use the more powerful existing tools, and why on earth would you make it public?
I will not claim to have read all the “tech experts“ posts. However I would like it they or even you could point to an example of good software that has never failed. Software can have flaws, but the spirit/goal of what is being presented by Apple is to make it without flaws and to fix them if they come up. I doubt they spent a few minutes writing some code without testing before they published their results. Thats just not how it works in big companies.
