[jimbokun]

LLMs sometimes spit out large chunks of text from it's training data almost verbatim.