Need to create a model to identify patterns in user details

I’m trying to create a model where I can detect fraudulent customers who try to access a certain app, using their personal details like fist name, last name, email and username. From the data that I have created, I’ve identified a pattern where the fraud customers use when they create their username or email along with their names, but I do that manually by going through the list. I need to create a model which identifies the pattern.

I decided to use NLP with character level tokenizing so that I could concatenate all four columns and try to identify a pattern with the characters since it cannot be done with word tokenizer.

Is there any suggestions or examples where I can refer for me to get a good approach ?