The other side of this is that your LLM interaction can be leveraged to train adversarial models that find ways to still deny or delay your claim despite your LLM-advised actions.

Insurance companies, or the companies they pay to launder their involvement, would pay a lot more for that than the public would be able to.