The Fight to Own AI-Generated Content is Just Starting

Last week, a Washington, D.C. judge ruled against computer scientist Stephen Thaler in his case against the United States Copyright Office (USCO), finding that works generated by an artificial intelligence (AI) system do not qualify for copyright protection.

In 2018, Thaler applied for a copyright covering the visual art piece A Recent Entrance to Paradise, the piece was created by an AI system Thaler developed and was made without any human input.

When Thaler filed for the registration, he listed the AI system as the author. The USCO denied the registration, saying that a work needed a human author to be protectable by copyright and, thus, receive a registration.

This prompted Thaler to sue the USCO. However, the judge in the case has dismissed Thaler’s arguments and granted a summary judgment motion in favor of the USCO. 

Thaler, for his part, has promised to appeal the decision. He claims that the court erred in its decision and failed to take into account many aspects of the case.

That said, Thaler lost an earlier case where he attempted to patent a piece of AI-generated code. In that case, the decision was upheld on appeal.

This new decision seems pretty decisive and the implications, at first blush, appear to be very clear: AI-generated works don’t qualify for copyright protection. 

However, Thaler’s work and these decisions deal with a very fringe scenario in AI, one that isn’t likely to apply to the vast majority of uses AI will have.

Not the Author

To date, most of the copyright issues around AI have focused on the way AI systems are trained and whether either that or the work it produces is copyright infringing.

The most prominent and public story about this was the recent class action lawsuit filed by comedian Sarah Silverman and others against OpenAI. However, it’s far from the only lawsuit filed as we’ve seen cases filed against GitHub, DeviantArt, Stability AI, Midjourney and more over the ingestion of copyright-protected work for the purpose of training AI systems. 

However, lurking in AI is a separate question: What, if any, rights and protections are there for works generated by AI systems? 

Thaler’s lawsuit(s) look to that question. However, his arguments look at the issue in an extreme way. Namely, Thaler isn’t listing himself as the author on either the patent or the copyright application, instead listing his AI as the author.

From a copyright standpoint, this outcome was wholly predictable. Back in 2018, we had a similar battle over a selfie taken by a crested macaque monkey. The USCO and the courts alike made it clear that the image was in the public domain as the monkey was the legal author and, since only humans can hold a copyright, no one holds those rights.

This means that any point Thaler is trying to make about the sentience or creativity of AI is moot. Even if an AI system were, without any doubt, a living and conscious being, it wouldn’t matter for the purpose of copyright as only humans can hold copyright in a work.

This helps copyright sidestep the philosophical issues of what is and is not alive and puts the focus on a much more straightforward question, and one that Thaler’s case doesn’t address: How much human needs to be in the work before a copyright can be granted.

The Other Problem with AI

Thaler’s lawsuit sits at an extreme fringe when it comes to how AI works are generated and used. Most who use AI don’t simply list their AI system as the author. Instead, they list themselves as the author, with varying degrees of veracity in that statement.

On one end, you have individuals who generate a work using AI and take full credit for it, without any indication of AI involvement. Ethically, this is clearly plagiarism, but it’s also legally dubious. The AI is the clear author in this case and, if the AI involvement is exposed, I wouldn’t expect them to qualify for any copyright protection.

At the other end would be creators that use AI to help generate ideas or outlines for a particular work that, ultimately, is built by a human. Here, the AI is only contributing non-protectable elements, so I would expect the human to enjoy full authorship and control of the world, both legally and ethically.

However, between those two extremes exists an entire spectrum of hybrid creative works. This includes AI-generated works that are heavily-edited by humans, human works supplemented with AI-generated elements, human-created worlds filled with AI characters, human books with AI illustrations and so on.

In some of these cases, it’s simple to draw the line between what is and is not protected, but as the more mixed the human and AI works get, the more challenging it becomes. At some point, a line has to be drawn when there is enough human for a work to qualify for protection and what that protection covers.

That, in turn, is going to be the real challenge in this space. However, it’s not a question that Thaler’s lawsuits even begin to address. As such, no matter how definitive those rulings seem, they don’t put the issue of copyright protection for AI-generated works to bed. 

It doesn’t even start to address the issues we’re likely to see at all.

Bottom Line

This ruling comes at a difficult time in the conversation around AI. Both the Writers Guild of America (WGA) and the Screen Actors Guild (SAG) are on strike, bringing the movie and television industry to a halt.

One of the key contentions between the unions and the industry has been the use of AI. The WGA and SAG are fighting to prevent the studios from using AI to either replace their work or reduce their pay. However, the studios call those demands unrealistic.

Some had hoped that this ruling might dissuade the film industry from pursuing AI. After all, copyright ownership is a key part of the film industry’s business model and the prospect of not having that ownership would be a death knell for AI in Hollywood.

However, Thaler’s cases are not illustrative of how studios would likely use AI. Studios still have plenty of reason to hope for and even expect that they would be granted control over films partially generated using AI.

In short, Thaler’s cases and this most recent ruling are not particularly helpful in the discussion about the role AI will play in the future of creative industries. Instead, they answer a question that has already been answered many times before.

Including once by a monkey.