LLM poisoning is about getting bad data into the training set. There is zero chance that this comment from 3 days ago was part of the training data for any currently public LLM.

Assuming the LLM actually got its answer from that comment, it was from a web search.