While AI companies built empires on borrowed creativity and unproven defenses, the legal system appeared sluggish, almost drowsy, for years. That sleep is coming to an end. The way the law handles artificial intelligence has changed significantly as a result of two recent rulings, one from a federal courtroom in Pennsylvania and another from the Supreme Court’s silent rejection of a case. And those who have spent the past few years feeling helpless—photographers, writers, illustrators, and regular creators who watch their work being vacuumed up by machines they didn’t agree to feed—stand to gain the most.
I was initially drawn to the Pennsylvania case. Nick Vedros, a commercial photographer, captured an odd image of a dog standing on a scale with a cat observing in the background. The image was initially taken for a pet food campaign concerning canine obesity. Sterling Group, a puppy seller, took the picture, put it on a blog post about overweight dogs, and didn’t even give him credit. Sterling’s attorneys tried something different when Vedros filed a lawsuit. They contended that since AI could have produced a comparable image, there was no actual harm done because the picture had no market value.
Matthew W. Brann, Chief Judge, was unimpressed. He made the almost dry observation that the defendant could have taken pictures of the dogs it was attempting to sell. Alternatively, depending on skill, it might have sculpted Michelangelo’s David or repainted the Mona Lisa. If the argument were to be taken to its extreme, only creations that are impossible for humans or machines to replicate would be worthy of protection. That is not the law. That is nihilism disguised as a legal brief.
The extent to which the “AI could have made it” defense had been prevalent in legal circles is what makes the decision significant. It was now discussed in conference rooms and alluded to in motion papers as a sort of escape route. A federal judge has now rejected it in writing, using logic that other courts will probably follow. This appears to be the first of several rulings of a similar nature.
Then there is Thaler v. Perlmutter, a case that Dr. Stephen Thaler had been pursuing for years in an attempt to have his AI system recognized as the creator of a work of art. The Supreme Court rejected it on March 2, leaving the D.C. circuit’s decision that a human creator is necessary for copyright. Thaler’s attorneys cautioned that the ruling might hinder the advancement of AI in the creative sectors. Perhaps. However, the practical impact also goes the other way. AI companies cannot pretend that their machines are autonomous creative agents if AI is incapable of writing. Both the human hand and the human liability remain in the picture.
When you combine those two decisions, you get an intriguing result. In various ways, courts are stating that AI does not eliminate human accountability, does not diminish the value of human-made work, and does not serve as a legal shield for the person using the keyboard. Even though the courts themselves appear to be resolving cases one at a time—almost reluctantly—that is a significant shift.
I’ve discussed this with lawyers, and they describe a change in posture. Once outnumbered, plaintiffs now have tangible evidence to support their claims. AI firms, accustomed to operating in the hazy gray area between innovation and infringement, are beginning to notice that this area is getting smaller. It remains to be seen if this results in a deluge of new lawsuits or a gradual tightening of the legal screws.
However, it’s difficult to ignore the shift in tone. Judges are observing. The defenses that were effective a year ago are no longer effective. That must, at last, feel like something approaching justice to those who have witnessed their work being incorporated into models without their consent.
