Except the list of hashes is a black box. Today it's child porn, tomorrow it's anti-government or something similar. It's a slippery slope that's only going in one direction.

Agreed, but as of today, we are at the top of the slope. Maybe we skip, maybe we don’t.

As implemented, I agree with Apple

Where there is a slope, riders will come. The only way to prevent people from riding the slope is to never built the slope in the first place.

“Anti-government” is absurdly vague. Not how this system works.

> “Anti-government” is absurdly vague.

That's the point. It's not an open to the public list, it's secret and controlled by few. It can contain whatever they want it to contain.

> Not how this system works.

That's absurdly naive.

The only naiveté is self-inflicted by reading clickbait articles instead of how the system actually works. When people start worrying about Apple scanning for abstract subjects like "anti-government" they're worrying about some hypothetical system, not the one that's been built.

No, we’re worried about the precedent it sets.

Scanning for CSAM on a server was a precedent for scanning on a device. Everything is a precedent for anything.

Your entire iPhone is a black box. If you don't trust Apple, there are so many points in the chain where they could backdoor you. The fact they are out in front of this is a good thing.

Also, I'm willing to sacrifice a tiny bit of privacy to stop child porn collections.

This[1] comment explains it very well. It just cements why I moved back to Linux on my laptop, and will be moving to something running open source software on my phone, and open hardware on all my devices.

[1]: https://news.ycombinator.com/item?id=28159353

But Apple doesn't report to authorities based only on a hash match.

They hash the files on your device.

The hashes are then sent out to for comparison to the hash table.

Right now they are only comparing to the CSAM table but they can add other later.

Also, how are the hashes sent? Can they be intercepted?

Yes but there is a manual review step. A hash match does not automatically report to authorities.

If the Apple reviewer looks at the images that were flagged and doesn’t see CSAM, they don’t report it.

A human at Apple reviews thumbnails of the flagged images to see if they are actually CSAM. You do not get reported just because of a hash match.

>Comparing hashes is not anything crazy.

If they just hash photos, and if they only use the NCMEC database and if they only look at photos being uploaded and if the government has no access, then it is not anything crazy. But by implementing the system they invite scope creep at every one of those ifs.

This is exactly how they describe it. If it changes, then I’ll change my position.