I'm just taking a stab, because my inorganic chemistry is rusty.
cyanate (CN-) is well known to bind to iron (it inhibits hemoglobin and causes anoxia), and it also forms a pretty strong complex with Zinc:
I would guess that in the case of the SOD that the other liganding amino acids and the pocket of the enzyme are such that the affinity for cyanide is lowered to the point that oxygen can displace it.
In the case of hemoglobin, CN will coordinate to the heme iron in much the same fashion that O2 does, but it doesn't come off. In the case of SOD, the enzyme will be looking to coordinate to superoxide O2(2-) which would be at a roughly tetrahedral angle to the metal. This is not a favored geometry for CN coordination, which prefers to linearly coordinate to things.
In the case CuZnSOD it seems likely that this enzyme has enough room and its Zn and Cu binding envirionments don't forbid CN binding. it would be an evolutionary accident really as to why one is sensitive and another is not.
some structures at rcsb:
thiocyanate bound. the pocket is pretty open between the two ions. no CN bound structure.