Post

Obviously, it’s quite difficult to opine on an opinion that hasn’t been published. But the news today that @ROSSIntel has lost its appeal in the third circuit does leave room for a couple of general observations regardless of the details.

First, as @AdamEisgrau has articulated, there are ways for the other AI cases to distinguish themselves from this one. The products are different. The process is different. The market dynamics are different. This is hardly a bellwether for the future of fair use in the domain of AI.

Second, Ross should’ve won. The raw material they copied in the process of training their AI tool wasn’t really expressive content at all.

West headnotes are just robotic restatements of public domain judicial opinions. They are the textual equivalent of the three-dimensional car scans in the meshwerks case.

The goal of headnotes is not to express anything at all, but rather to capture and reproduce as accurately as possible, already existing public domain expressions, facts and ideas.

This case should have been lost on copyrightability before it ever got to the question of fair use. Third, the outcome is obviously terrible policy.

It leaves in place a toxic oligopoly of legal research titans, who have built an empire by locking up the law and selling it back to us.

The kind of competition that Ross would have brought to the marketplace is exactly the competition that Copyright law seeks to foster through doctrines like the fact-expression, and idea-expression dichotomies, and of course, through fair use.

Creating new useful tools that make public domain information more publicly accessible is the kind of thing praised by courts in cases like Feist and Google Books.

We will see what the court’s opinion actually says by way of justification, but there is really nothing they can say that would render this outcome reasonable under current copyright doctrine, or desirable as a policy outcome.