Update: This paper is now being considered for retraction according to this editorial note. The Scientist has just reported criticisms of the paper also, and the article appears with a critical comment on pubmed.
Considering the myriad of applications that CRISPR-Cas9 can and will be in employed I think research like this certainly is important. However, the sample size (N=2) is far too small to draw any conclusions. I'm surprised that a Nature paper think a sample size like that is sufficient, for any type of analysis. However, the observation of recurrent off target events is puzzling. The sensation around this paper is too much, definitely, but perhaps it's good to get everyone more alert about what possible side effects the "silver bullet" CRISPR-Cas9 might have. Follow-up of this work is required.
So the message of this paper is not appropriate, but so are Chinese scientists who already use this immature technology on human subjects. I guess this is just how science works - different findings (some more pronounced than others) after which a consensus is found.