It's one thing to enable a viable birth for a child that would otherwise be 100% guaranteed to have a life-limiting genetic disease, but modifying a gene- a gene we do not know that much about- on the chance of reducing a 5% risk of contracting a virus that can be effectively treated- is pretty well beyond what most people would consider ethical.
To put it another way: there would be a 95% chance any CRISPR modifications would have not prevented the baby contracting HIV because 95 out of 100 such babies wouldn't have contracted HIV anyway. And those 95 babies (and at least some of their children) would live with that genetic variant forever.
If the baby was going to be nonviable and you had the chance to eliminate the disease by knocking out a faulty gene or replacing it with the version of the gene everyone else has, that's one thing. But this is a rare gene variant. There may be a reason why only a small proportion of people have it! It could well be that the side-effects of lifetime ART are actually better than the unknown side-effects of this gene.