By Gary Symons
TLL Editor in Chief
The US Copyright Office says works created with the help of artificial intelligence may be copyrighted, in a policy with major implications for the licensing industry.
From now on, all copyright applicants are legally required to disclose when their work includes AI-generated material. As well, any previously filed application that does not disclose the use of generative AI must be corrected, or copyright could be withdrawn.
However, what really has the lawyers talking is the new guidance on what can and cannot be protected.
Prior to the guidance issued on Wednesday, copyrights could only be issued for images, videos, or stories created by human beings. The new policy from the US Copyright Office says a copyright can be issued if a work is created that combines the work of an AI like ChatGPT with the work of a human being. “In each case, what matters is the extent to which the human had creative control over the work’s expression and actually formed the traditional elements of authorship,” the office said.
The guidance document was issued not long after the US Copyright Office considered for the first time whether a work created largely by the generative AI bot MidJourney could be issued a copyright.
“The Office reviewed a registration for a work containing human-authored elements combined with AI-generated images,” the Office said in its policy statement. “In February 2023, the Office concluded that a graphic novel comprised of human-authored text combined with images generated by the AI service Midjourney constituted a copyrightable work, but that the individual images themselves could not be protected by copyright.”
In that case, the Copyright Office rejected parts of a claim for copyright protection for a graphic novel entitled Zarya of the Dawn, by Kris Kashtanova. (Shown in photo at top) However, Kashtanova viewed the ruling as a partial victory, because the Office did grant protection for some parts of her claim.
Kashtanova used the “generative AI” program MidJourney to produce a series of images which were used in the graphic novel alongside her human generated storyline.
Initially the Copyright Office granted protection to the comic book in its entirety, including the images, but in a follow-up review the Office learned that an AI program had generated the images. As a result, the Copyright Office granted protection for the story itself, but not for the images in the story.
Kashtanova still feels that the images should also be covered, since they were generated by her commands to the AI, and they also were generated in such a way as to fit in to the story she wrote.
But the Copyright Office disagrees, saying, “In the Office’s view, it is well established that copyright can protect only material that is the product of human creativity. Most fundamentally, the term ‘‘author,’’ which is used in both the Constitution and the Copyright Act, excludes non-humans.”
Intellectual property lawyer Ryan Meyer of the firm Dorsey & Whitney says the policy guidance statement is not surprising to see, because the Copyright Office has actually been wrestling with the issue for the past five years.
“The Copyright Office’s policy statement is not hypothetical or premature,” Meyer said. “Starting in 2018, examiners have received multiple applications for works authored by AI. The policy statement represents the Copyright Office’s recognition of an escalating trend.”
Meyer also says the ruling in the Zarya of the Dawn case is not the end of the debate over whether works created with the help of artificial intelligence can be copyrighted. The key, he says, is how much input the human creator has over the final product of the generative AI program.
“Consistent with past court and Copyright Office decisions, the policy statement requires a work to be the product of a human author,” Meyer points out. “However, the policy statement clarifies that a human author might create a work using AI as an assisting instrument, like a computer, Adobe Photoshop, or guitar pedals]. Going forward, examiners must determine, for each case, whether a human author used an AI like a tool to form the author’s own original expression or whether the work represents the AI’s expression.”
The Copyright Office itself uses the example of a camera, used by humans to capture a still or moving image. Obviously, the human did not create the image that is being captured, but the Copyright Office does grant protection to original photos because humans decide the composition of the photo or video.
That principle came up in a leading case on intellectual property, Burrow-Giles Lithographic Co. v. Sarony, in which a defendant was accused of making unauthorized copies of photographs. The defendant argued photographs could not be copyrighted because they were created by a camera, not by a human being, but the court disagreed.
In its ruling, the court said photos “… are representatives of original intellectual conceptions of the author.’’ As well, the court defined an ‘‘author’’ as ‘‘he to whom anything owes its origin; originator; maker; one who completes a work of science or literature.”
And that’s where the grey line currently sits in the debate over creative works that are produced through the use of generative AI. Those who argue that such works should enjoy copyright protection say generative AI is legally no different than a camera, which is the device that actually captures the image.
But Meyer says it’s not the device used that matters; it’s the degree to which the creative process is driven by a human mind.
“Elements of a work created by feeding prompts to a generative AI will likely not be registrable under the new policy for lack of human authorship,” he explains. “The policy statement compares such prompts to instructions given to a commissioned artist who decides how to express the instructions in the final product.”
AI Cannot Get Copyright Protection, But Neither Can Monkeys
A case in point would be the Copyright Office’s example of a photo taken by a monkey, which was featured in another US IP law case, in which the court held that “a monkey cannot register a copyright in photos it captures with a camera because the Copyright Act refers to an author’s ‘‘children,’’ ‘‘widow,’’ ‘‘grandchildren,’’ and ‘’widower;’’ terms that ‘‘all imply humanity and necessarily exclude animals.’’
A monkey with a good lawyer might argue that it can have monkey children, but the point is that the US Copyright Office only issues copyright protection in which an image, video or other work is created under the supervision of a guiding mind.
In the case of Zarya of the Dawn, the images were created by a generative AI, although they were done using the prompts from the human author. Right now, however, it appears that won’t count with the Copyright Office, which will continue to require that a human being has to wield the paint brush, or the mouse, while the image is being created.
The trick for creators using AI in the future will be in exercising as much control as possible over the composition of the image, how the image looks, and so on, to convince the Office inspectors that a human mind conceived of the image, and the AI only physically created it similar to the way a camera captures an image that is composed by a photographer.
“Applicants who use AI as part of their creative process should realize that their registration might be restricted to the parts of the work authored solely by a human,” says Meyer. “For elements of the work created using AI-assistance, the applicant should explain how the human author, not the AI, exercised the traditional elements of authorship, such as expression, selection, and arrangement.”
Genius Brands Is Probing the Limits of Copyright Protection for AI-Assisted Works
This is far from an academic exercise for those working in the arts, which also will include large film studios and virtually anyone else who creates texts or images. As TLL reported in February, Genius Brands announced it is creating the first-ever children’s series produced by artificial intelligence, called Kidaverse Fast Facts.
The series creators are using OpenAI’s ChatGPT program, as well as other generative AI technologies. Genius is making the move in order to speed production of its various series, to cut costs, and increase its overall efficiency.
“AI is a game changer for us, providing significant cost efficiencies and speed to market, allowing us to expand our content initiatives and fuel our programming pipeline in a way that couldn’t have been possible before,” said Genius Networks president Todd Steinman. “These AI tools and the content they create will certainly have a huge impact in children’s entertainment and beyond. With its innovative use of AI, we are confident that Kidaverse Fast Facts will be a hit with kids and parents around the world!”
However, the guidance document issued by the US Copyright Office may raise questions for Genius Brands and for any other studio using AI. For example, if Walt Disney had used AI to create the iconic Mickey Mouse character, then the image of Mickey Mouse could not be protected by copyright. Rather, a human would have to illustrate Mickey Mouse, while AI might be used to generate the background, to animate the character, and so on.
Similarly, if ChatGPT is used to write all or most of the story and dialogue, then that too would not comply with the Copyright Office’s new rules. For any company following this route, the balancing act will be to ensure that it can show enough of the creation is done, or at least strictly supervised, by human beings.
Lawyers Evan Gourvitz and Lara Ameri wrote in a legal briefing in March that copyright eligibility will be defined by the balance between the human and AI contribution.
“For example, when an AI program produces a complex written, visual, or musical work in response to a prompt from a human, the “traditional elements of authorship” are determined and executed by the technology—not the human user—and, thus, the resulting work is not copyrightable,” the lawyers pointed out. “On the other hand, a work containing AI-generated material may be copyrightable where there is sufficient human authorship, such as when a human selects or arranges AI-generated material in a creative way or modifies material originally generated by AI technology.
“Ultimately, copyright protection will depend on whether the AI’s contributions are “the result of mechanical reproduction,” or they reflect the author’s “own mental conception,” the Copyright Office said. “The answer will depend on the circumstances, particularly how the AI tool operates and how it was used to create the final work.”
Netflix and Coca-Cola Seek a Path Forward for AI-Generated Work
Another company that has kicked the hornet’s nest of generative AI is Netflix, as it debuted the animated short Dog and Boy in January. Directed by Ryotaro Makihara, the story follows a young boy and his robotic dog as they make their way through a post-apocalyptic world.
In this case, Makihara and Netflix did not create or animate the characters using AI, but they did use AI to create the background scenes.
That appears to be the type of use case the US Copyright Office has in mind when it considers a work to be eligible for copyright protection. Dog and Boy’s story line and its characters are both created by human beings, which can be copyrighted, while the background typically would not need to be protected by copyright.
But it’s not just artists and entertainment creators who are testing the boundaries of copyright in the use of AI.
A marketing campaign from Coca-Cola called “Create Real Magic” combines OpenAI’s GPT-4, the startup’s most advanced large language model to date, and DALL-E, which generates images based on text prompts.
Coca-Cola is opening the platform to digital creatives around the world, with the idea that it can ‘democratize’ its brand iconography and “highest-profile advertising assets,” Pratik Thakar, global head of creative strategy and integrated content for the Coca-Cola Trademark, stated in a recent announcement. Thakar said the experiment is meant to demonstrate Coke’s commitment to testing and scaling AI technology quickly.
As part of the program, Coke selected four AI artists in different markets, including Emma Sofija from the US, Chris Branch and Paul Parsons from Europe, and Ean Hwa Huag in Asia to help kick off the campaign. Coke is also launching a workshop program for another 30 creatives who will get to use the tools to co-create content that could be leveraged for licensed merchandising, digital collectibles and more. Participants will be credited for their work, Coke says, but the question is whether the works can actually be copyrighted.
Again, that will depend on how the art is created, and if the US Copyright Office decides it does not include enough human guidance, it’s possible Coca-Cola will not be able to copyright any of the images to be used in this campaign.
The permutations of the policy are obviously complex, so the Copyright Office will be organizing a series of seminars on the topic for creatives and companies.
Throughout the spring, the Office will host public listening sessions with artists, creative industries, AI developers and researchers, and lawyers working on these issues. These roundtable-format listening sessions will provide an opportunity for participants to discuss their goals and concerns related to the use and impact of generative AI in creative fields.
Later this year, the Office plans to publish a notice of inquiry soliciting public comments on a wide range of copyright issues arising from the use of AI. The seminars will cover four different types of creative work as they apply to the AI policy, with the first starting in April.
Interested parties can register for the public listening sessions using the links below:
- Literary Works on Wednesday, April 19, from 1:00 p.m. to 4:00 p.m. eastern time
- Visual Works on Tuesday, May 2, from 1:00 p.m. to 4:00 p.m. eastern time
- Audiovisual Works on Wednesday, May 17, from 1:00 p.m. to 4:00 p.m. eastern time
- Music and Sound Recordings on Wednesday, May 31, from 1:00 p.m. to 4:00 p.m. eastern time
The Office says it may not be able to accommodate all requests to participate in the listening sessions, but will provide other opportunities for interested parties to share their views on AI policy with the Office.
As well, following the listening sessions, the Office says it will continue engaging with the public through informational webinars during the summer. In support of this initiative, the Office has also launched a new webpage for announcements, events, and resources related to AI and copyright at copyright.gov/ai.
While the future may see more changes in policy, for now it appears that only works created primarily by humans will be eligible for copyright protection, and any companies who wish to benefit from that protection will need to ensure they can prove that imagery, music and stories are primarily created and designed by the human mind. Sorry robots!