While unsafe memory access can have consequences for stability, it has much broader implications than that. A system can keep running reliably while being heavily exploited.
I am really tired of this FUD. Any system can be like this. You can have perfectly memory safe language and still be compromised. If you have enough dosh you hire experts and have them try to poke holes in your systems or analyze whether those have been compromised already. Until then all this security talk is not worth much.
But those would be logic errors, not memory management errors. If someone is stupid enough to pass user input directly to a shell or database then yes, they will get compromised.
But these comprise only 30% of all security errors as the research shows. Most are due to memory management issues.
I don't see how it's FUD. Yes, any system can have security problems, but some are more likely to than others. "Just try harder" is no solution. With equal effort, better (suitable) tools still give better results.
Let's just disagree. I think your definition of "better" is vastly different from mine. I also think that a mess most use as a current web development stack using "safe languages" poses way more danger security wise then my proprietary client/servers ever will.