When you code, you're building on top of assembly / cpu registers and instructions. The fact that X86/64 runs on a bunch of electrical fluctuations is actually inconsequential and an implementation detail.
Any x86/64 processor built of transistors naturally needs to take into account the behavior of the transistors it's made up of. But once you've gone above the micro-code/processor level, those details are fully abstracted.
Regardless what programming language you use, that is abstracted away identically and matters not.
However, the fact that registers in the process and memory locations exist and mutate is not something abstracted away from every language. It's abstracted in some, but not others, so it's valid to say that this difference is within the scope of the programming language.
That is why the argument differs. If the above isn't clear, I'd be happy to give another go at explaining it.