I think validations must be shared by all the Use Cases that use the Entity. If a Use Case doesn’t do the same validations than another, we risk having inconsistent data.

I think the validations are a part of the Entity identity. For example, the User you define in your article isn’t just a data with random fields like first_name, last_name, etc. These fields are constrained by validations that define the concept of User (like “the first_name field must be longer than 2 characters”).

Also all the Use Cases using this Entity have to do the same validations, so we must at all costs avoid repetition.