In a galaxy not so far away, specifically, the Middle District of Pennsylvania, a federal court delivered the first ruling I could find rejecting a seemingly appealing argument for copyright defendants: the idea that a work should lose its copyright protection because artificial intelligence could hypothetically have created something similar. In Vedros v. Sterling Group of the Twin Tiers, Inc., Chief Judge Matthew W. Brann quickly put that argument to rest. 

What happened?

Nick Vedros, a commercial photographer known for his humorous, animal-based advertising photography brought a copyright infringement claim based on his photograph depicting a dog placing its front paws on a scale while a cat looks on nearby - a creative composition originally produced for a dog food advertising campaign targeting overweight dogs. 

The defendant, Sterling Group, sells puppies. It copied the entire photo, without alteration or credit, and displayed it as the header image above a blog article on its website entitled "A Breeder's Note on Canine Obesity.” Sterling Group didn’t deny an infringement occurred. They instead raised a fair use affirmative defense under 17 U.S.C. § 107 (see "When standards become law, fair use kicks in. Implications of recent Third Circuit decision in ASTM v. UpCodes" for a fair use refresher). 

On the fourth fair use factor, the effect on the market for the copyrighted work, the defendant advanced a novel argument: that because a similar photo could have been generated using artificial intelligence, the original lacked any meaningful commercial value and so could not suffer cognizable market harm. 

You heard that right. So did Chief Judge Brann who quickly dispatched the idea. The court observed that the defendant could have taken its own photograph of its dogs - the same dogs it sells. Taking the point further, the court noted that the defendant might just as well have sculpted “Michelangelo’s David” or repainted the “Mona Lisa,” depending on its talent. Taken to its logical conclusion, the defendant’s position would mean that only works no human or machine could recreate would be entitled to protection. 

Ultimately, the court granted summary judgment in favor of Vedros on both the copyright infringement claim and the fair use defense. 

So what?

There is a lot of ongoing conversation around the question of whether AI-generated outputs themselves can receive copyright protection, a question addressed in Thaler v. Perlmutter and by the U.S. Copyright Office's evolving guidance. But Vedros addresses the mirror-image question: Can a defendant escape liability for copying a human-created work by arguing that AI could hypothetically have produced something similar? The answer as you might have guessed is a resounding no. 

The force may not be with those who copy first and rationalize later. Courts are watching, and they are not impressed by the AI defense.