Scenario A: A program has a memory vulnerability. That's bad, and the reason it's bad: attacker-controlled data can potentially trigger this vulnerability, which could even lead to privilege escalation.

Scenario B: A program writes machine code in an executable region of memory, and the code has a memory vulnerability. That's bad, and the reason it's bad: attacker-controlled data can potentially trigger this vulnerability, which could even lead to privilege escalation.

Scenario C: A program write machine code to disk, which is then read into memory and executed. That's bad, and the reason it's bad: attacker-controlled data can potentially trigger this vulnerability, which could even lead to privilege escalation.