This particular data set is fine, but we've seen other people release bigger data sets thinking it was anonymised only to find that it wasn't. (See, for example, the Netflix data dump.)
Most people don't care about their movie rentals, but will be a lot more cautious about some of their medical history.
I'm in the UK. Rules here are pretty strict. Mostly that's a good thing; you run your intended research by a research review panel, and if it needs ethics approval you do that too. The benefits of that are that people get help from a real mathematician early in the project design so they should be getting the stats and the sample sizes etc right.
Like I said, I'm only gently concerned. And I'm sure they'll get this right.
Most people don't care about their movie rentals, but will be a lot more cautious about some of their medical history.
I'm in the UK. Rules here are pretty strict. Mostly that's a good thing; you run your intended research by a research review panel, and if it needs ethics approval you do that too. The benefits of that are that people get help from a real mathematician early in the project design so they should be getting the stats and the sample sizes etc right.
Like I said, I'm only gently concerned. And I'm sure they'll get this right.