Public discussion of artificial intelligence and copyright has focused on two main issues: training data and output.

Lawsuits have raised, and courts have begun to answer the questions (1) whether using copyrighted works as training data constitutes fair use and (2) under what circumstances artificial intelligence infringes copyright in training data.

In this article, the rules of copyright law are applied to try to answer the question whether an artificial intelligence itself might be copyrightable subject matter.