Besides advertising their bots as slaves, they may just be using prompts as training data (or the whole chat histories in reinforcement learning), in which case they probably would prefer you don’t swear.