@b3ll real-life opposite example;
Me: “if I use X, will Y data be safe?”
Agent: “yes, X will ignore data it isn’t aware of”
X: “DROP TABLE Y”