Is there any evidence showing that copy number variation changes over time? I'm wanting to model interactions in expression level as a dynamic bayesian network, but an assumption my approach will need to make is that it is static.
Your question could be phrased more specifically to avoid ambiguity, but rephrasing it the way that I suspect you mean it, ("Is there any evidence showing that [the rate of] copy number variation changes over time?"), then yes, there is indeed.
The rate depends on many factors including which mechanism and which organism and which region of the organism's genome and as @Michael wrote, also what scale (et. al.) is under consideration too.
So your assumption that the rate is static should probably be stated explicitly.