http://www.srm.org.uk/papers/CBlocks3D_programmable_construc...
http://www.srm.org.uk/papers/CBlocks3D_programmable_construc...
http://www.complex-systems.com/pdf/04-6-2.pdf
The two problems with 3D life have always been
1) 3D is computationally very expensive compared to 2D, so you tend to get stuck working with much smaller grids; and
2) you need a really good visualization system, otherwise interesting patterns and uninteresting patterns both look like ugly random blobs of cells and you can't tell what's going on.
That said -- not promising anything, and especially not on any definite timeline, but the next version of Golly might have a Lua-based system for experimenting with 3D CAs.