I would suggest that you are not the first person to see this, as it is a fairly common issue across many datasets. For taxa or regions that have fewer identifiers, mis-identifications can indeed snowball and create issues. It is a problem with Millipedes for example. I’ve seen it to a smaller scale with bees (a lot Pyrobombus), and it’s one of the reasons why we tend to look so carefully at out of range observations.
Best solution is to place the proper ID’s, a lot of them.