[shittyadmin]

I'd say that the sandboxing introduced by mobile OSes today solves the vast majority of the problem.

By isolating applications and introducing permissions, malware that can steal or encrypt user data isn't possible even for people installing those pirated APKs.

[vbezhenar]

Don't trust Android or iOS or macOS sandbox. Google invests huge amounts of development time and research into Google Chrome JavaScript sandbox. It is the real wild west, there are malicious actors who want to break that sandbox. There are multiple layers of protection. Yet there are successful attacks. Much less people trying to break Android or iOS sandbox, because you can just ask for permission from Android and because Apple can kick bad app from the store and prevent infestation. It means that security of those sandboxes is worse, there are many undiscovered (or undisclosed) holes.

Check out history of Java sandbox with its numerous vulnerabilities. I have no reasons to expect anything different from built-in sandboxes. It's like relying on unix user permissions and allow to run anything under untrusted user. Works in theory, but you'll be owned pretty soon, because local root escalation vulnerabilities are not that rare.

In those days the only sandbox I would trust is JavaScript one. It's battle tested.

[shittyadmin]

> because you can just ask for permission from Android

You cannot ask for permission to bypass sandbox restrictions on Android. You need root access, which means physical access to do things like unlock the bootloader or an exploit.

iOS sandbox seems slightly weaker here due to the use of hidden/private functions to protect certain things, sideloaded apps would likely be a bigger risk on iOS than Android at the moment, but that's not something unresolvable.

In any case, the things you're discussing aren't really so problematic - isolation systems are only getting better, OS level ones are improving every day. We could easily have sandboxes at this level just as secure as the javascript ones.

[comex]

But the type of malware described in the article obviously is possible, given that it’s possible even on the Play Store. The OP posed switching to a decentralized model as a solution to that, and it’s hard to see how that makes any sense.

[shittyadmin]

I'd argue the type of malware described in this article is fairly to completely harmless to the user. Harm to ad networks and a bit of wasted bandwidth is basically the worst case scenario.

The wasted bandwidth would be made clear by the OS to the user too, so it'd be trivial to identify if it was a significant consumer.

[raxxorrax]

But in the end the software running on our phones is mostly crap.

I doubt user data being protected by these mechanisms helped people to guard their data.