Yes, there's a lot of disagreement about policies regarding model opensourcing especially. It seems likely to me that some employees at large AI labs (Google Brain Deepmind, Meta, Microsoft Research, etc.) will always disagree with the overall policy of their organisation. This creates a higher base rate risk of insider threats.