Sounds like maybe a GLOVAP-based change. Don't know of anyone else supporting that change (yet), so I say yes, deviate.

do you have an example of a POWO url with an accepted Horkelia or Horkeliella species?

As for the deviation question, it may prove true that these genera (Ivesia, Horkelia, Horkeliella, maybe others) make Potentilla paraphyletic. But I am not aware of recent work supporting that conclusion, so I don't feel that there is (yet) any basis for departing from currently accepted North American taxonomy for these genera. Once this is studied, it's possible that the decision would be to further split up Potentilla s.str., rather than make it even more heterogeneous than it already is.

I could copy and paste a similar comment on many of the other recent POWO lumps of North American genera.

If there was a deviation in place for Horkelia and Horkeliella, it's gone now, FYI

This doesn't appear to be a completely unwarranted change, as a wider concept of Potentilla that includes Potentilla sensu stricto and any member of the Alba clade (e.g. Potentilla alba) appears to necessarily include Ivesia, Horkelia, Horkeliella, Potentillopsis, and Tridophyllum to be monophyletic, especially if it includes the Argentea clade. Eriksson, Persson, &. Smedmark (2022) list a number of supporting literature sources that state that this lump of the Ivesioid clade is stable and well-supported, but that a wider lump including the Anserina clade (Argentina and Tylosperma) would be unstable. They choose to support a wider concept of Potentilla.

POWO has a tendency to be super lumpy with large, complicated groups (pteridophytes are the best example), so that appears to be their approach with this broad concept of Potentilla. Personally, I would prefer the Ivesioid clade remain distinct, but we may run into difficulties keeping the Alba/Argentea clades out of Potentilla. Here's what Eriksson et al. (2022) say about it, with some edits for brevity and clarity in this context:

"It is fully possible to rank the Reptans clade as a genus... Classifying [the Reptans clade] as Potentilla would be disruptive for species names... The main drawback would be that all the species in the clade that contains most Potentilla species as currently classified (clade G, the Argentea clade, ca. 350 species) would need to be transferred to another genus (the currently monotypic Potentillopsis)... it would cause complications due to the fact that there is evidence that there has been gene flow between the Reptans clade and both the Alba clade and [the Ivesioid and Argentea clades], in some cases having given rise to species through hybridisation (Persson & al., 2020a; Persson & al., in review). Also, the secondary criterion of stability of names argues against ranking the Reptans clade as a genus, regardless of whether its name would be Potentilla or Tormentilla."