To me, the main benefit is potential knowledge about specific risks that I might be able to mitigate. But even if it just satisfies my own intellectual curiosi

To me, the main benefit is potential knowledge about specific risks that I might be able to mitigate. But even if it just satisfies my own intellectual curiosity, I find it to be a benefit. To the world, I think that having a large dataset helps to improve the quality of studies such as the one being discussed. There's great utility in having a "hold-out set" for verification. The OP presumably perceives some similar benefits, hence the sense of being "conflicted". Whether the benefits are real is a valid, although separate question than whether the privacy concerns are justified.