Yes, species/lineage/population distinctions are quite fuzzy at the level of divergence under consideration here (dire wolves vs gray wolves).

Here's what was actually have done according to the New Yorker article, starting with a gray wolf genome as the baseline:

After almost a year of computational genetic analysis, Colossal researchers used Crispr to make twenty edits on fourteen genes. Fifteen edits were derived from Colossal’s study of the dire-wolf genome and five tweaks were derived from scrutiny of the gray-wolf genome.

20 edits and 14 genes -- clearly some related to coat color, however:

But the genes that guided coat color presented a problem: they carried with them a risk of blindness and deafness. (In humans, variations of these genes can lead to Waardenburg syndrome, which causes pigmentation deficiencies, among other problems.) So the group decided to edit a different gene that, when expressed in dogs, also codes for a lighter coat.

So the coat color alleles are NOT the dire wolf alleles.