[AnthonyMouse]

I guess at that point you're basically asking whether it's possible to make higher resolution printers than cameras, but considering you can in principle do printing using lithography similar to what they use to make semiconductors, that's probably going to win over the average phone camera. Although you're obviously then talking about a much more sophisticated attack.

[mjg59]

It's not just a matter of printing, it's a matter of placement. If you can carry equipment of that calibre into a hotel room and do the swap then that'll defeat things, but it's not clear that that's realistic.

[AnthonyMouse]

You wouldn't necessary need it to be in the hotel room. You sneak in, take a picture, have the lab down the street reproduce it, come back in a half hour and make the swap.

That's also assuming you would actually need that level of sophistication. It's plausible that there is a level of printing technology somewhere between "crappy inkjet" and "semiconductor fab clean room" that could still fool a phone camera.

There is also the possibility of accessing the inside of the machine without tearing the sticker. You think they're going to disassemble it by removing the screws, but they actually disassemble it by slicing off a section of the case with a sharp blade and then epoxying it back together. Or make their modifications through the cooling vents.

And that's really the other problem too. If you don't know how they're going to do it, you don't know what to look for to detect that they did. Your sticker is intact so you're safe, right? Right?

[mjg59]

You're still left with needing perfect placement, which isn't something that's realistic to do by hand. Physical case modifications are also going to be detected by any reasonable tooling (there's at least one vendor who can tell you which physical mold something came off on the production line via phone camera imaging, they're definitely going to spot a glued together hole in the case). So you're left with going in via existing case holes as the most realistic option, which has raised the bar by a significant amount - this is now an attack that's going to take much longer and require a higher level of skill, so the probability that it'll be carried out is reduced by a lot.

Nobody is realistically going to say that a computer plugged into the internet is unhackable. Instead the goal is to make it sufficiently difficult to hack that it's either cheaper to solve the problem a different way or target a different person. The same is true here. Nobody believes it's literally impossible to compromise an iPhone when you have physical access, but it's considered hard enough that almost any other option is preferable. We should be holding laptops to the same standard.