That is absolutely true. But but you can write memory-bug-free code in Zig but you cannot prevent heap allocations in most of the languages listed in the article, making it outright impossible to write certain software in them.
Sure one can write memory-bug-free code in x86 assembly too. But how can you prove it? ATS is an example of a low-level systems language where you can prove it.
A function that doesn't allocate doesn't mean it's safe. (Indeed if anything the opposite is more likely - copying is safe, if slow, writing to something that was passed in tends to be what breaks).
formalized subsets of x86 assembly exist. Coq can be used as a macro assembler. tools to work with llvm ir exist, x86 can be raised up to llvm ir and proved, kinda bad way tho.
