>researchers say LLMs rarely spit out training data verbatim unless interacted with adversarially, but theoretically, they could. Theoretically they can genera

>researchers say LLMs rarely spit out training data verbatim unless interacted with adversarially, but theoretically, they could. Theoretically they can generate any arbitrary snippet of code (if it correctly fits the distribution), regardless of whether or not the code was in the training dataset.