/s?
"I have a criticism" -> <criticism> makes it more <good/trustable/etc> not less
phrasing is a classic AI hallmark, doubling down on nonsense via the shortest path which is plausibly maybe correct sometimes. In the context of a broken AI rewrite, that felt a bit like an intentional piece of humour, enough to warrant asking.
Beyond that, in the context of a linker, userspace logic matters much more than generic "memory safety" or whatever. The core algorithm is dynamically choosing which bytes to load and execute, and being off by a byte as a userspace bug is a major safety failing, regardless of whether it's "memory safe" or has "fearless concurrency" or other such bullshit. I'm not aware of any study, or even any good anecdotes, suggesting that Rust is better on average from a userspace logic pespective than even Python of all things. In the context of a linker, I somewhat doubt that there's any evidence of Rust having "lower rates of exploitable vulnerabilities than C or C++ code." It might be better, but I doubt that evidence exists yet.
Lol. You've got AI paranoia dude. It's pretty obvious that Rust is more reliable than C++ in many ways, and trying to blame such an absolutely obvious statement on "AI" just because you disagree with facts to try to discredit this statement just makes you look silly.
> In the context of a linker, I somewhat doubt that there's any evidence of Rust having "lower rates of exploitable vulnerabilities than C or C++ code."
That it's a linker does mean exploits would be a bit more special case than if it was say a web server yes. But the number of exploitable issues is still going to be way higher, even if the real world severity is low.