It's time for U.S. Copyright officials to come to terms with AI-generated art
Should we really be relying on legal precedent from the 1800s to decide whether work generated by AI today should have copyright protection?
AI-scientist Stephen Thaler has traveled the world seeking intellectual property rights on behalf of an AI program he wrote. To clarify, Thaler has asked several jurisdictions to credit his AI – dubbed DABUS – as the inventor for devices it created.
He hasn’t had much success.
Bureaucrats in Australia, Germany, the EU, the U.K., and the U.S. have denied his applications. He has appeals in most of those places and more applications pending in a dozen of other countries.
His shining moment came in 2021 when South Africa’s Patent Office granted DABUS a patent for a food container it invented.
You might be tempted to raise a skeptical brow as one TechDirt writer did in describing Thaler as the “dude who keeps suing.” But that misses the point.
These are test cases. Thaler is working with pro bono attorneys at The Artificial Inventor Project to promote “dialogue about the social, economic, and legal impact of frontier technologies such as AI and to generate stakeholder guidance on the protectability of AI-generated output.”
Thaler, president and CEO of Imagination Engines, and The Artificial Inventor Project lawyers aren’t advocating on the behalf of robots – they aren’t trying to argue that AI is a sentient being with inherent and inalienable rights.
They are trying to advance the rights of people who are using AI to create things. That would include myself and, presumably, many of the folks who read this newsletter.
Thaler has another chance to make his case here in Washington, D.C. where he sued in June, challenging the U.S. Copyright Office’s long-held stance that only work created by a human can be copyrighted.
Thaler sued after the U.S. Copyright Office denied copyright protection for the computer-generated image shown above, which he titled “A Recent Entrance to Paradise.” Thaler applied for copyright protection in November 2018 but the U.S. Copyright Office rejected his application in August 2019 because the work lacked “human authorship.” Two subsequent requests for reconsideration were also rejected.
This week, Thaler asked a federal judge for summary judgement, basically saying there’s no need to go to trial because the facts are not in dispute and the law supports his position.
Thaler’s brief (which you can read here) covers a lot of bases in making a case that the U.S. Copyright Office should not have denied his registration. (In a companion case, a federal appeals court in October rejected Thaler’s attempt to list his AI system as the sole inventor on a patent application.)
Patent attorney Ryan Abbott, who represents Thaler in both the patent and copyright cases, told Reuters that a decade from now it will "seem odd that people were wondering if we should protect these sorts of things."
Stuck in the past
One point that Thaler’s legal team makes is that the U.S. Copyright Office is way behind the times.
In rejecting Thaler’s application, the U.S. Copyright Office cites legal precedent and policies that were created long before anyone envisioned AI generating the type of work we’re seeing today.
Incredibly, the legal debate over AI-generated work raises some of the same questions courts wrestled with when asked in the 19th Century if copyright law was intended to protect photographs, or in the 20 Century when they had to decide if the protections applied to moving pictures.
Kicking the can down the road
Anyone who has spent time generating complex images on Midjourney or Stable Diffusion knows that humans are an integral part of the creative process. Yes, the AI generates the image, but a human comes up with the artistic concept, formulates the descriptive text prompt and then revises the text any number of times in order to shape the results.
It’s a human-machine collaboration, at the least. However, the slippery slope for copyright officials comes in quantifying how much the human contributed and whether the contribution was enough for a human to claim authorship.
Many of the AI-creators I’ve talked to see AI as a tool that they use to create art – not unlike the camera that a photographer uses. They feel a sense of accomplishment, if not ownership, for the images they generate.
Rather than figuring out how to adapt to new technology and new ways of thinking about art, the U.S. Copyright Office’s intractable position is more of a “kick the can down the road” approach. Let the next person figure it out.
For a brief moment in the fall, it seemed that maybe the U.S. Copyright Office had finally come to terms with technology. In a surprise move, the office granted a copyright registration to AI-generated art for Zarya of the Dawn, a comic book created by New York-based photographer and UX designer Kris Kashtanova.
And why shouldn’t Zarya have copyright protection? Kashtanova conceived the storyline, wrote the text and dialogue and directed Midjourney’s AI in creating the images.
However, a few weeks after registering Kashtanova’s copyright in September, the U.S. Copyright Office notified Kashtanova that it had made a mistake and that Kashtanova could lose their copyright protection because AI-generated art was used to illustrate the comic book. Kashtanova’s attorney filed a response on Nov. 21 and they are awaiting the U.S. Copyright Office’s next move.
A question of ownership
Interestingly enough, Thaler and his attorneys don’t seem to be making the AI-is-just-a-tool-used-by-artists argument. Instead, they say the work, in and of itself, is original and worthy of copyright, even if AI, and not Thaler, was the creator.
Rather, Thaler is the “owner” of the art and that gives him the right to copyright it. He developed and owns the AI that created the art and likens the situation to a “work-for-hire” arrangement where the copyright belongs to the person who hired the creator. This is an arrangement that media outlets and companies often make with photographers and illustrators.
One final point to keep in mind is that Thaler tried to register his work in 2018, well before the current explosion of AI-art (Thaler’s work was actually created a couple of years earlier). He was truly out there by himself tilting at windmills.
In 2022, it’s estimated that people around the world used AI to generate more than a billion images. That’s a billion images out there with unclear provenance.
For argument’s sake, say that only 1,000 images (a ridiculously low number) out of the billion generated last year meet some hypothetical threshold for human authorship – that’s still a large enough number to deserve attention from copyright officials. (We can debate the artistic merits of AI-art but copyright doesn’t require a qualitative test, just that the work be original i.e. not copied.)
If our copyright laws are intended to encourage artists to create works that benefit the public good, while also protecting their economic interests, then withholding potential copyright protection from millions of creators is leaving a hell of a lot of truly artistic work in the lurch.
It’s time for the U.S. to bring its copyright law into the 21st Century.
All images – except were noted – were generated by Joe Newman using the Midjourney platform. You can also follow along on Instagram at @most_sublime_media. Get fresh content delivered to your inbox by subscribing to our Substack newsletter.