Weekly take: Why there’s hope for AI users after Midjourney ruling

For those who think intellectual property should be reserved for humans, the US Copyright Office’s decision to deny protection to artificial intelligence-generated images may seem like a win.

But AI users and enthusiasts shouldn’t lose heart.

In fact, the decision has several positive takeaways for those calling for a greater AI influence.

The US Copyright Office’s decision, handed down on February 21, concerned the graphic novel ‘Zarya of the Dawn’, created by artist Kristina Kashtanova.

Kashtanova created the text herself but used AI technology Midjourney to assist with the images.

According to the US Copyright Office's ruling, only the human-produced elements of the novel and the compilation of the texts and images were worthy of copyright protection.

The decision should somewhat settle the position on whether AI-generated work is copyrightable, at least in the US. That is, of course, assuming Kashtanova decides against an appeal.

The ruling as it stands also offers guidance for other jurisdictions that may look to the US for precedents on issues concerning AI authorship.

This would mean fewer disasters from IP offices when faced with new technology. They will want to avoid a similar problem to India’s IP office (IPO), which in 2021 threatened to cancel a copyright registration that recognised an AI tool as co-author despite granting registration a year earlier.

The IPO has kept quiet since the owner of the AI tool in question, Ankit Sahni, responded in December 2021 that the office didn’t have the power to cancel the registration.

Kashtanova’s registration, coincidentally, followed a similar journey to Sahni’s.

After granting the registration in September 2022, the US Copyright Office backtracked in October that year when it initiated a cancellation action against the registration.

The office's latest ruling, which limited the registration, came after Kashtanova argued that even though she used Midjourney to help her create some of the images, the assistance it provided didn’t diminish her contribution.

Stakeholders could find themselves supporting or opposing the ruling depending on their interests.

But the decision at least offers some guidance and sets the ground for further debate.

A major positive for AI users is that the copyright office upheld the copyright registration in the graphic novel, accepting Kashtanova’s argument that the use of the Midjourney tool didn’t diminish the human mind that conceived and created the final work.

The office also recognised that where Kashtanova had made substantive edits to an intermediate image generated by Midjourney, those changes could provide human authorship and would not be excluded from protection.

The observations mean a copyrightable work containing content generated by an AI tool could still be protectable, so long as it comprised some human-contributed element.

Of course, much would depend on the extent of the contribution by the AI system and the human creator, but the observation can be considered a small victory for AI enthusiasts.

It’s also noteworthy that the copyright office's decision to deny protection to the images generated solely by Midjourney was based on how that particular tool operated.

In its refusal, the office ruled that Midjourney users couldn’t be authors because they didn't control the tool and because it was impossible to predict what the system would create ahead of time.

One could assume that if another AI tool operated differently from Midjourney – for example, if it required considerably more human intervention to generate the images, or if its results were predictable – then such output could be worthy of copyright protection.

And, of course, there is no guarantee that the US Copyright Office will blindly follow the 'Zarya of the Dawn' ruling for all AI-generated works submitted before it.

Finally, the most crucial aspect stakeholders must remember is that the decision only interprets US copyright law as it currently stands.

The US, according to a notice published by the USPTO on February 14, has already begun deliberations on the future of AI and inventorship with regard to patents.

The USPTO is seeking input on several issues surrounding AI inventorship, including what statutory changes should be considered to inventorship law and whether AI systems should be made eligible to be listed as an inventor.

The office’s consultation also indicates that allowing AI systems to be listed as inventors could possibly promote and incentivise innovation.

Stakeholders expect the office to initiate a similar consultation on copyright law and AI before long.

It may be that patents come first but it might not be long before copyright law is revised to account for new technologies.

Crucially, any changes to the law will also need to balance the burgeoning role of AI against the belief that IP protection be retained solely for humans.