(1) whether they're too sneaky with how they tell consumers that their data will be used for training
(2) whether they are creating hallucinated libelous/harmful content about individuals online.