[Question] Multiple class decision trees

Hey everybody, I have a question about using decision trees for multiclass classification.

I'm familiar with the concept of a "One-vs-all" approach where for N classes you would learn N models, where the ith model represents if an instance is classified as the ith class or not. This may seem silly, but my question is what do you do after you've trained those models and want to classify a new instance? What if more than one of your models say that this instance belongs in the ith class?

Thanks for reading