I'm pretty sure most people have thought about doing this; the issue is that with our current technology, the overhead is too large to be practical. At least, t

I'm pretty sure most people have thought about doing this; the issue is that with our current technology, the overhead is too large to be practical. At least, that was my understanding in the matter when I was doing some cryptography related research. On the other hand, a lot of research is striving towards that goal. A method for fully homomorphic encryption was very recently discovered, and although it is still way too slow to be practical, the fact that we finally figured out a way to do it after several decades of searching is a huge step.