Sometimes the partially valid backup contains all that is really needed to recover, the parts that change rapidly may not be "valuable", in the sense that restoring an older copy of some files in the archive may not be a breaking issue, vs not having any data at all.
The ability to keep a backup that technically isn't complete, but is complete enough, is better than nothing. Chances are, even if the backup doesn't verify, it's still usable enough for disaster recovery, especially if you keep multiple dates of backups.
Some (many?) containers have the option to keep internal database backups, and since those files are pretty much guaranteed to be stable even when the container is still running and changing the active database, the "invalid" backup still contains a valid backup made by the app itself that can be used.
Making backups of running containers is complicated, it's not a black and white issue like your statement seems to imply.