As you're reviewing your errors, a user can in the case they see a false positive mark the error as "not an error", this does 2 things:
* Reports it back to the AI as training data to help improve accuracy
* Removes it from the error list of the user so that they no longer see the mistake.