ETA: I was off by an order of magnitude (originally I said sub-1%) because the doping is extensive, and the mass of the lead/gold is a dominant fraction of the total. The formula given in the paper is (subscripts in brackets):
Pb[10−x]Cu[x](PO4)[6]O with 0.9 < x < 1.1.
Similarly for Au we would have Pb[10−x]Au[x](PO4)[6]O. Taking the centerpoint x=1, this becomes Pb[9]Au[1](PO4)[6]O. In other words, there would be one gold atom for every 9 lead atoms.
The "unit cell weight" is 2647.59 g/mol, and the molar mass of gold is 196.97 g/mol, so in fact the hypothetical gold weight content is about 7.44%, not sub-1%.
That said -- presumably superconducting transmission wires would be thinner than the ones we are used to (a function of critical current rather than resistance). So I'm not sure that we'd have a theft problem worse than we already have with copper.