How can it be guranteed that the statstical model commonly known as ai was not trained on leaked photoshop code? If that has happened then it wouldn't be a clean room implementation.