Calling it a surname (the intended meaning) instead of a "last name" (synonymous with the intended meaning in certain limited cultural contexts) would solve part of the problem, but as Gaile said, it would be akin to a Band-Aid on an artery.
The problem starts with basing a profile's unique database identifier on user input. This is bad database design, because it leads to all sorts of problems when that user input turns out to be wrong. I can appreciate wanting something name-like rather than the alphanumeric soup used by other sites (such as FamilySearch), but it doesn't really work -- I can't keep track of which number goes with which ancestor, even when there are only two of them with that surname. This means that I have to look the person up by something other than the DB ID and then copy-and-paste the ID, exactly like I do with the gibberish strings on FamilySearch.
There has been a proposal to redesign the entire name field structure of the database to allow for all known variations (and then some): https://www.wikitree.com/g2g/536471/db-schema-expansion-name-table
I have no idea whether there are any plans to implement this idea, and if yes, how long that may take.
In the meantime, the database's name structure is definitely a major flaw of WikiTree.