Yes, there are both scientific and ethical problems. The scientific problems have to do with the representativeness (or otherwise) of the limited number of SNPs (single nucleotide polymorphisms) included in the 23andMe genotypes. This is not whole-genome analysis, of course. It is for such reasons of scientific accuracy that the FDA says it has now limited the company's operations. The ethical ones are really more to do with consent to specific uses of the data and ownership than with confidentiality. Some participants in the research arm of 23andMe (called 23andWe) have criticised the company's downstream use of their epidemiological and genetic data. They feel that they contributed the former altruistically and paid to have the latter analysed, but the firm is making profits from their contribution, which they made altruistically, by selling it on to pharmaceutical firms and other commercial companies. They argue that they did not give consent to these commercial uses. You might like to refer to my book on personalised medicine, Me Medicine vs. We Medicine, chapters 3 and 7 (Columbia, 2013).
Thanks for your thoughtful and detailed responses. I will certainly get a copy of your book and read the relevant chapters. The benefit/risk ratio gets more complicated with more thought!