What’s happened?
The United States Copyright Office (USCO) has released a report outlining its approach to granting copyrights to content created using generative AI.
The verdict? A work created solely through AI is not copyrightable, but a work that combines human creativity with AI can be copyrighted, so long as there is a “sufficient” amount of human expression in that work.
“The use of AI tools to assist rather than stand in for human creativity does not affect the availability of copyright protection for the output,” the USCO report stated.
The Copyright Office concluded that, at least on this particular AI issue, no new laws are needed, as existing precedent can guide the granting of copyrights to works involving AI.
“Questions of copyrightability and AI can be resolved pursuant to existing law, without the need for legislative change,” states the report, which can be read in full here.
“After considering the extensive public comments and the current state of technological development, our conclusions turn on the centrality of human creativity to copyright,” said Shira Perlmutter, Register of Copyrights and Director of the USCO.
“Where that creativity is expressed through the use of AI systems, it continues to enjoy protection. Extending protection to material whose expressive elements are determined by a machine, however, would undermine rather than further the constitutional goals of copyright.”
The report is in keeping with the office’s approach thus far. For instance, in 2023, the USCO was sued by computer scientist Stephen Thaler, who was denied copyright on content created by his DABUS AI system.
In that instance, the court sided with the USCO, declaring that a work that was created entirely by AI can’t be copyrighted. Thaler is appealing the decision.
Nonetheless, for creators and copyright owners, the USCO’s rules offer a measure of stability: It looks almost certain that fully AI-generated works – for instance, the songs created by AI music-making platforms like Suno and Udio – won’t enjoy copyright protections.
The USCO report gained immediate support from the Recording Industry Association of America (RIAA).
“RIAA applauds the US Copyright Office for reaffirming the longstanding tenet of copyright: human authorship is required in order for a work to be copyrightable,” the group said in a statement.
“That foundational principle is the cornerstone of a pro-innovation future and essential for human creators. The irreplaceable contributions of creators are essential to maintaining and growing America’s world-leading intellectual property economy.”
“RIAA applauds the US Copyright Office for reaffirming the longstanding tenet of copyright: human authorship is required in order for a work to be copyrightable.”
RIAA
However, things get a little more complicated when it comes to AI-assisted creation. That’s a relevant issue for the many artists and producers who have begun using AI tools to help create music.
The USCO notes that “whether human contributions to AI-generated outputs are sufficient to constitute authorship must be analyzed on a case-by-case basis.”
All the same, the USCO has offered enough detail to determine, in a broad sense, what kind of work will be covered, and what won’t, in the age of generative AI.
When is AI-enabled content copyrightable?
There are, of course, many different ways to use AI to create music and other works. The USCO report addressed three common ways AI is used today:
- Using prompts to get a particular result from a generative AI engine
- Inputting a human-created work, and asking the AI to alter it in some way
- Modifying or arranging AI-generated content
Prompts:
Eliciting AI-generated works using a prompt – usually, text typed into the AI engine – is the most clear-cut case here: You can’t copyright a work created just by entering a prompt.
The USCO report offers a number of reasons for this, one key one being that prompts are ideas, not expressive creations, and ideas can’t be copyrighted. If your only contribution to a creation is an unprotectable idea, you can’t claim copyright.
Also, the USCO says that prompts don’t give the human user enough control over the outcome to claim authorship.
“The output of current generative AI systems may include content that was not specified and exclude content that was,” the report states. “The fact that identical prompts can generate multiple different outputs further indicates a lack of human control.”
But what about “prompt engineering”? Some AI users have developed a skill out of repeatedly entering prompts into an AI engine, altering their request one prompt at a time until they get the desired result. Should that output count as a copyrightable work?
Again, the Copyright Office says no. Citing copyright law precedent, the report says that “selection of a single output is not itself a creative act.”
Borrowing an analogy from one of the thousands of submissions it received from the public on this issue, the USCO quotes:
“If I walk into a gallery or shop that specializes in African savanna paintings or pictures because I am looking for a specific idea (say, an elephant at sunset, with trees in the distance), I may find a painting or picture that fits my idea, [but that] in no way makes me an author.”
However, the USCO notes that “there may come a time when prompts can sufficiently control expressive elements in AI-generated outputs to reflect human authorship. If further advances in technology provide users with increased control over those expressive elements, a different conclusion may be called for.”
Expressive inputs:
Another way to use AI is to input something you have created and have the AI enhance it. For instance, a visual artist could input a sketch and ask the AI to turn it into a photorealistic 3D image. Or, a musician could input a melody they composed and have the AI turn it into a full “recording” with various instruments and a backbeat composed by the AI.
In this instance, the USCO says only those parts of the work that a person created can be copyrighted; those parts created by AI cannot.
“Where a human inputs their own copyrightable work and that work is perceptible in the output, they will be the author of at least that portion of the output,” the report states.
“Where… creativity is expressed through the use of AI systems, it continues to enjoy protection. Extending protection to material whose expressive elements are determined by a machine, however, would undermine rather than further the constitutional goals of copyright.”
Shira Perlmutter, USCO
Granting copyrights in these instances will be guided by the existing rules for copyright in “derivative works,” i.e. original works created on top of some other pre-existing work. For instance, a remix.
In a derivative work, only the new, original aspects can be copyrighted by the person who created it. If a producer creates a remix, they can copyright the new backbeat they added, but not any elements of the original song.
A copyright in this case “may also cover the selection, coordination, and arrangement of the human-authored and AI-generated material, even though it would not extend to the AI-generated elements standing alone,” the USCO says.
Modifying or arranging AI-generated content:
Let’s imagine a situation where an artist uses AI to create a number of guitar lines, bass lines, and drum beats. They then rearrange those elements to create a new track. Is that copyrightable?
The USCO says yes – but only if the human creator has modified those elements “in a sufficiently creative way.”
But, as with expressive inputs, only those elements that the human creator is responsible for would be copyrightable. In the case of this hypothetical track, it would mean that the arrangement of sounds is copyrighted, but the individual, AI-created sounds aren’t.
“The inclusion of elements of AI-generated content in a larger human-authored work does not affect the copyrightability of the larger human-authored work as a whole,” the USCO says.
“For example, a film that includes AI-generated special effects or background artwork is copyrightable, even if the AI effects and artwork separately are not.”
What’s next?
This USCO report is the second in a series of three that lay out policy on copyright and AI.
The first report, released in July 2024, addressed the issue of AI-generated digital replicas (i.e., deepfakes), and concluded that “a new law is needed” to protect people and copyright holders in an era when it’s supremely easy to replicate a person’s face, voice or other attributes. Congress is currently deliberating a number of such proposed laws.
But the third report, which the USCO says is “forthcoming,” may be the one that the music industry and other copyright holders will be watching most closely.
That report will issue a verdict on using copyrighted material to train AI. For IP-dependent industries, this has become among the largest (if not the single largest) issue with AI.
Rightsholders have launched numerous lawsuits against AI developers who they believe used their copyrighted materials without permission or payment.
In the music industry, the most closely watched lawsuits these days are those launched by the majors against Suno and Udio, two upstart music creation platforms that Sony Music, Universal Music, and Warner Music say trained their AI on copyrighted music.
Suno and Udio plan to defend themselves by arguing that they should be granted a “fair use” exemption under US copyright law; the music companies disagree.
The USCO’s approach to this could guide the courts in deciding whether or not AI companies should be held liable for copyright infringement – or whether they will be able to continue using copyrighted content in training their AI models. That’s a scenario that many in the music industry see as apocalyptic, with the potential to destroy human creativity and the music industry with it.
A final thought
The USCO report offers some clear wins for human creators and copyright holders – most notably, with its clear rejection of copyright for fully AI-created works.
Nonetheless, the other principles set out by the USCO could complicate matters for rights holders going forward – though, to be fair, it’s these new technologies that are complicating things, not the USCO’s rules.
For one thing, determining copyright infringement could get trickier. A recording owner may recognize that another creator ripped off an element of their recording, but their copyright claim will be rejected because the element was created by AI.
For this reason, the USCO has issued registration guidance requiring copyright petitioners to disclose how they used AI to create their work. Yet the rule creates an incentive to hide the use of AI – if for no other reason than to protect as much of a recording or composition as possible from copycats.
It may yet turn out that we do need some new regulations or guardrails to ensure creators aren’t cheating the system, and copyrighting AI-created elements.
From a commercial standpoint, for the music industry, uncopyrightable song elements could reduce the licensing opportunities from remixes.
Similarly, the above-mentioned movie studio may use AI to generate the background scenery in a movie, only to find itself without recourse if that scenery appears in another studio’s movie.
Users of AI, beware.Music Business Worldwide
‘ The preceding article may include information circulated by third parties ’
‘ Some details of this article were extracted from the following source www.musicbusinessworldwide.com ’
