Let's play with two examples of 'touching'

Sample A: Simple
You have alu and copper in the same block and it's assembled with metal bolts/screws. So long as the alu and the copper both touch the electrolyte (the water), the circuit is complete and galvanic corrosion will occur.

Sample B: Elaborate
Let's say you have a GTZ with the LGA1366 backplate and an aluminum radiator. If they're connected via rubber/plastic tubing, you're alright....until you mount it in a case. At that point, you have the alu radiator touching the case, the case touching its own motherboard tray, the LGA1366 backplate touching the motherboard tray, the thumbscrews screws touching the backplate, the mounting plate touching the thumbscrews, the assembly screws touching the mounting plate, and the copper base touching the assembly screws. In this situation, your alu rad and copper plate are electrically connected, there's an electrolyte, and galvanic corrosion will occur!

There are three ways to avoid galvanic corrosion:

1) don't mix metals (silver and copper counts as mixing, so does alu and copper, so does stainless and silver, so does stainless and copper, so does XXXXXXX and YYYYYY, even a small difference in nobility will lead to galvanic corrosion over time [nobility of metals such as alu and SS change a lot based on pH, so they're harder to look up, I'm not sure what they are--but silver and copper are surprisingly far apart]). This is practically impossible for us considering how important copper is to our thermal goals and the use of stainless steel in our favorite pumps (DDC and D5). Even using a killcoil 'breaks' this.

2) don't use an electrolyte. Also impossible...this is liquid cooling. There are basically no liquids that will not act as an electrolyte (and none that are worth our time, thermally). EDIT: there are actually many liquids that won't act as an electrolyte, distilled included, but they'll all eventually ionize over time (months to years), thus acting as an electrolyte.

3) don't complete the circuit, electrically. This is the only one we have some control over. We can use only plastic barbs, delrin tops, coated backplates, etc. It also largely invalidates our inherent mixed metals in #1 (the stainless steel in Laing pumps is not in circuit, electrically, neither is a kill coil in a reservoir or tubing).

Here's a list of blocks that I have in front of me that have a 'galvanic circuit' (multiple wetted metals in electrical contact) when used as intended (using included barbs counts):
Heatkiller 3.0 LT/Cu (LC does not, LT+Silicone mod does not either)
Apogee XT
Apogee GTZ SE
Sapphire Rev.A
Luna Rev.A
Aqua Computers Cuplex XT di
Aqua Computers Cuplex XT di2 (I don't actually have this one here, but it has the same fundamental construction as the original di)
Koolance CPU-350 (acrylic top)
Koolance CPU-345 (acrylic top)
EK Supreme
Alphacool Livingstone
Alphacool Yellowstone

I'm sure there are more on the market too...as well as many GPU blocks with dissimilar metals in direct contact with each other. Anything with a metal top will almost definitely fall into this category (using a metal barb instantly includes it). Of course, the less dissimilar the metal, the slower and less noticeable the g.c. will be, to the point where it's basically unnoticeable on the timescales we use our hardware.

All in all, the galvanic corrosion thing is way overblown (especially here at XS). Most of the times we've 'seen it' it's really something else at play that's gunking things up

Also of note, oxidation (i.e., rust and the other metal's versions of it) is not the same thing as galvanic corrosion!