Ok, I might have been to hasty in commenting on your last recap. Your baseline is sound. In any case, we're talking about a medical help/ advice tool. If it's not providing any value, any interaction with it (let alone sharing medical data) is pointless and a waste of time. So I think any convincing argument against sharing private data with it should take in consideration at least a minimum of potentially missed valuable information. Otherwise it's an easy argument to make, but also an empty one.

In this case, I suspect that the classic biases of HN (pro-privacy and anti-ai) might interact to dismiss the value that can be provided by a specialized medical llm/ agent (despite indications that an unspecialised one is already helpful!) while rightly pointing out the risks of sharing sensitive data.