How to Own AI-Generated Content? A Dive into Copyright Law

Introduction

In the digital age, where artificial intelligence (AI) has become a significant player in content creation, understanding the intricacies of copyright law is essential. This article will delve into the complex world of copyright law, derivative works, transformative works, and how they relate to AI-generated content.

The advent of AI has brought about a paradigm shift in various sectors, including the creative industry. From writing articles and composing music to creating visual art, AI has demonstrated its potential to generate content that was once solely the domain of human creativity. However, this has also led to a host of legal and ethical questions, chief among them being – who owns the copyright to AI-generated content?

As we navigate through this uncharted territory, it becomes crucial to understand the fundamental principles of copyright law and how they apply to the rapidly evolving landscape of AI-generated content.

What Do You Get With a Copyright, Exactly?

Copyright law grants creators exclusive rights to their original works. These rights include the ability to reproduce the work, create derivative works, distribute copies, and perform or display the work publicly. In essence, copyright law seeks to protect the expression of an idea rather than the idea itself.

For instance, if you write a novel, you cannot copyright the idea of a love story set in medieval times. However, you can copyright the specific expression of that idea – the unique plot, characters, and dialogues that make your novel distinct.

It’s important to note that these rights are not absolute and are subject to certain limitations and exceptions, such as fair use. Fair use is a legal doctrine that promotes freedom of expression by permitting the unlicensed use of copyright-protected works in certain circumstances. These circumstances can include criticism, comment, news reporting, teaching, scholarship, and research.

Understanding the nuances of copyright law becomes even more critical in the context of AI-generated content. If an AI program writes an article or composes a piece of music, who owns the copyright to that work? Is it the developer who created the AI program? Or is it the user who provided the inputs and parameters for the AI to generate the content? Or does the work fall into the public domain because it lacks human authorship?

These questions highlight the legal complexities surrounding AI-generated content. As we delve deeper into this topic, we will explore derivative works, transformative works, and the role of human creativity in copyright law. We will also examine recent court cases and legal rulings that shed light on these issues.

In doing so, let’s aim to provide a comprehensive understanding of copyright law as it applies to AI-generated content, offering insights for creators, users, and legal professionals navigating this exciting and challenging new frontier.

Derivative and Transformative Works

Derivative works are new, original products that include aspects of a previously copyrighted work. For instance, a movie based on a book, a translation of a novel into another language, or a remix of a song are all examples of derivative works. The key aspect of derivative works is that they incorporate a significant portion of an existing copyrighted work, but also add new, original elements that make the derivative work distinct.

However, creating a derivative work requires permission from the copyright holder of the original work. Without such permission, creating a derivative work could constitute copyright infringement. This is an important consideration for creators using AI tools that might draw upon existing copyrighted works to generate new content.

Transformative works, on the other hand, add something new or alter the original work in a way that gives it a new meaning, message, or purpose. A transformative work is not merely a reproduction of a copyrighted work, but rather a new creation that comments on, criticizes, or provides a different perspective on the original work.

The concept of transformative works is closely tied to the doctrine of fair use, which allows for the use of copyrighted material without permission in certain circumstances. Courts determine whether a work is transformative by considering whether it adds new expression or meaning to the original work, and whether it uses the original work for a different purpose.

Both derivative and transformative works can be protected under copyright law, provided they add significant creative elements to the original work. However, the line between derivative and transformative works can be blurry, and determining whether a work is derivative or transformative often requires a case-by-case analysis.

The Human Element in Copyright Law

The U.S. Copyright Office clarified that only humans can own copyright following the infamous ‘monkey selfie’ case. In this case, a macaque monkey took a selfie using a camera belonging to photographer David Slater. The photo went viral, and a dispute arose over who owned the copyright – the monkey, the photographer, or no one?

The U.S. Copyright Office ruled that a photograph taken by a monkey could not be copyrighted, stating that copyright law only protects “the fruits of intellectual labor” that “are founded in the creative powers of the mind.” Because the monkey was not legally capable of having a creative mind, it could not own the copyright.

This means that for a work to be copyrighted, there must be an element of human creativity involved. Courts have ruled that the bar for minimum creativity is low, especially in the realm of exact science and factual texts. Even a small amount of creativity, as long as it is original and comes from a human, can be sufficient for copyright protection.

This raises interesting questions in the context of AI-generated content. If an AI creates a work independently, without human intervention, can that work be copyrighted? If so, who owns the copyright? These are complex issues that are still being explored in legal circles, and the answers could have significant implications for the future of AI and creative work.

The Low Bar of Human Creativity

The concept of creativity is central to copyright law. For a work to be eligible for copyright protection, it must be original and exhibit a modicum of creativity. However, the threshold for what constitutes creativity is surprisingly low. This is particularly evident in the realm of appropriation art, where artists use pre-existing objects or artworks to create new works, often with minimal alterations, and get copyright to it.

One of the most notable figures in the world of appropriation art is Richard Prince, an American painter and photographer. Prince is known for his “rephotography” technique, where he photographs existing photographs and presents them as his own work. His art has sparked numerous legal battles over copyright infringement, raising questions about the nature of creativity and originality.

In one of his most controversial works, Prince took screenshots of other people’s Instagram posts, added his own comments, and sold the prints for hefty sums. In 2014, he faced a lawsuit from photographer Donald Graham, who claimed that Prince had used one of his photographs without permission. Prince’s defense was that his use of the photograph was transformative and thus fell under fair use.

The court ruled in favor of Prince, stating that his work, while minimally transformative, had a different aesthetic and conveyed a different message than the original photograph. This case highlighted the low bar for creativity in copyright law, where even a small alteration to an existing work can be enough to constitute a new, copyrightable work.

In the context of AI image generation, such as with models like Midjourney and DALL-E, Prince’s case could have interesting implications. It would mean that, if your additions are at least as creative as Mr. Prince’s, then you can own copyright to the whole AI image. Also, since the Ai models were trained on existing copyrighted images as part of their input and the copyright holders have initiated class action lawsuits, it could potentially be argued that this falls under fair use, similar to Prince’s defense, because Midjourney definitely adds more creativity than him.

However, it’s important to note that the transformation needs to add a new expression, meaning, or message to the original work. The changes would need to exhibit a degree of creativity and add a new layer of meaning or interpretation to the image. This sounds difficult but it really isn’t as you can see in Mr. Prince’s transformative masterpieces.

Another one of Prince’s most controversial works involved using images from a book by Patrick Cariou, ‘Yes Rasta’, for his collage paintings. Prince scanned several of Cariou’s images into his computer, printed them onto his canvases, and made limited alterations. This sparked a copyright infringement lawsuit against Prince, which resulted in a U.S. district court ruling against him and the Gagosian gallery that exhibited his work.

The low threshold for creativity is not limited to the art world. In the realm of text output, for example, the arrangement of facts in a phone directory was deemed creative enough to warrant copyright protection in the case of ‘Feist Publications, Inc., v. Rural Telephone Service Co.’. The Supreme Court ruled that even though the individual facts (names, phone numbers) were not copyrightable, the selection and arrangement of those facts could be, as long as they exhibited a minimum degree of creativity. In the context of AI text output, this case suggests that the selection and arrangement of facts could be considered a creative act. If an AI, like ChatGPT, generates a unique arrangement of facts or information, that output could potentially be copyrightable. However, simply changing the order of facts in the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.

Another text output example is the case of ‘Matthew Bender & Co. v. West Publishing Co.’ involved the pagination of legal reports. West Publishing Co. argued that their unique pagination system was a creative act deserving of copyright protection. The court agreed, stating that although the creativity involved was minimal, it was enough to warrant copyright protection. This case implies that the unique arrangement of text, such as a pagination system, could be considered a creative act. If an AI generates a unique arrangement of text, that output could potentially be copyrightable. However, simply changing the pagination of the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.

The Case of ‘CCC Information Services, Inc. v. Maclean Hunter Market Reports, Inc.’: In the context of AI text output, this case suggests that the selection and arrangement of data could be considered a creative act. If an AI, like ChatGPT, generates a unique arrangement of data or information, that output could potentially be copyrightable. However, simply changing a few words or phrases in the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.

The Case of ‘Harper & Row, Publishers, Inc. v. Nation Enterprises’: This case implies that the selection and arrangement of quotes or excerpts could be considered a creative act. If an AI generates a unique selection and arrangement of quotes or excerpts, that output could potentially be copyrightable. However, simply rearranging the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.

The Case of ‘Nichols v. Universal Pictures Corp.’: This case suggests that the specific expression of ideas, such as the way a story is told, could be considered a creative act. If an AI generates a unique story or narrative, that output could potentially be copyrightable. However, simply changing a few details in the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.

The Case of ‘Baker v. Selden’: This case implies that the specific expression of a system or method could be considered a creative act. If an AI generates a unique explanation or illustration of a system or method, that output could potentially be copyrightable. However, simply rewording the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.

Despite the scarcity of AI cases, the US copyright law’s requirement for minimal human creativity in various fields implies that copyrighting AI-generated content may not necessitate extensive human involvement either. This situation creates a dual scenario of opportunities and challenges. On the positive side, even slight human input could render a massive AI-generated creation (say, an entire novel or your own version of Wikipedia) eligible for copyright protection. Conversely, the risk of unintentional copyright infringement escalates due to the potential similarity in output generated by AI tools.

In all these cases, it’s important to note that the bar for creativity is low, but not non-existent. Simply changing a few words or details may not be enough to claim copyright over an AI’s output. The changes would need to exhibit a degree of creativity, such as a unique selection and arrangement of data or a unique expression of ideas.

As AI continues to evolve and become more integrated into the creative process, the legal landscape will need to adapt. Clear guidelines will be needed to determine what level of human input is necessary for an AI-generated work to be considered original and creative. Until then, creators will need to tread carefully, ensuring that their use of AI respects existing copyright laws while pushing the boundaries of what is possible.

Tips on Modifying AI Output to Own It

To own AI-generated output, you must first ensure that the AI’s Terms of Use grant you a license to the output. Then, you must add a small element of human creativity to it. This could be as simple as tweaking the AI’s parameters, adding a unique title, or combining multiple AI-generated pieces into a new, original work.

Modifying AI output to meet the minimum required level of human creativity can be achieved in several ways. For instance, you could add a unique commentary or critique to AI-generated text, incorporate AI-generated artwork into a larger design, or use AI-generated music as a backdrop for a human-performed voiceover.

Adding Annotations or Explanations

Adding annotations or explanations to AI-generated content can be a way to infuse human creativity. This could involve providing additional context, explaining complex concepts, or offering your own interpretation of the AI’s output. By adding your own insights and understanding, you are contributing a creative element that goes beyond the AI’s output.

Incorporating AI Output into Human-Created Content

You can also incorporate AI-generated content into human-created content. For example, you could use AI-generated data or graphics in a research paper or report, or use AI-generated music or sound effects in a film or video game. By integrating the AI’s output into your own original work, you are adding a layer of human creativity that could potentially make the work eligible for copyright protection.

Reformatting or Reorganizing AI Output

Reformatting or reorganizing AI-generated content can also be a way to add human creativity. This could involve rearranging the sections of an AI-generated article, changing the layout of an AI-generated infographic, or remixing an AI-generated song. By actively shaping the structure and presentation of the AI’s output, you are adding a creative element that goes beyond the AI’s output.

Tweaking AI Parameters

One of the simpler ways to add a touch of human creativity to AI-generated content is by tweaking the AI’s parameters. This could involve adjusting the AI’s settings to produce a specific type of output, such as changing the style or tone of AI-generated text, or altering the color palette or composition of AI-generated artwork. By actively shaping the AI’s output, you are adding a layer of human creativity that could potentially make the work eligible for copyright protection.

Combining Multiple AI-Generated Pieces

Combining multiple AI-generated pieces into a new, original work is another way to add human creativity. This could involve merging several AI-generated images to create a unique collage, or stitching together multiple AI-generated text snippets to form a cohesive article or story. By actively selecting and arranging the AI’s output, you are adding a layer of human creativity that could potentially make the work eligible for copyright protection.

Adding Unique Commentary or Critique

Adding unique commentary or critique to AI-generated content is another way to add human creativity. This could involve providing your own insights or analysis on an AI-generated article, or offering a critical perspective on an AI-generated artwork or piece of music. By adding your own thoughts and ideas to the AI’s output, you are adding a creative element that goes beyond the AI’s output, potentially making the work eligible for copyright protection.

Incorporating AI-Generated Artwork into a Larger Design

Incorporating AI-generated artwork into a larger design is another way to add human creativity. This could involve using an AI-generated image as a background for a website, or incorporating AI-generated graphics into a logo or brand identity. By actively shaping and integrating the AI’s output into your own design, you are adding a layer of human creativity that could potentially make the work eligible for copyright protection.

Using AI-Generated Music as a Backdrop for a Human-Performed Voiceover

Using AI-generated music as a backdrop for a human-performed voiceover is another way to add human creativity. This could involve using an AI-generated soundtrack for a podcast or video, with a human voice providing narration or commentary. By combining the AI’s output with your own performance, you are adding a creative element that goes beyond the AI’s output, potentially making the work eligible for copyright protection.

In conclusion, while the legal landscape surrounding AI-generated content is still evolving, there are several strategies you can use to add a touch of human creativity to AI-generated output. By doing so, you can potentially secure copyright protection for your work, opening up new possibilities for creativity and innovation in the digital age.

FAQ

What is the difference between derivative and transformative works?

Derivative works and transformative works are two distinct concepts in copyright law. A derivative work is a new, original product that includes aspects of a previously copyrighted work. Examples of derivative works include translations of a novel into another language, a movie based on a book, or a remix of a song. These works incorporate a significant portion of an existing copyrighted work but also add new, original elements that make the derivative work distinct. However, creating a derivative work requires permission from the copyright holder of the original work. Without such permission, creating a derivative work could constitute copyright infringement. This is an important consideration for creators using AI tools that might draw upon existing copyrighted works to generate new content.

On the other hand, transformative works add something new or alter the original work in a way that gives it a new meaning, message, or purpose. A transformative work is not merely a reproduction of a copyrighted work, but rather a new creation that comments on, criticizes, or provides a different perspective on the original work. The concept of transformative works is closely tied to the doctrine of fair use, which allows for the use of copyrighted material without permission in certain circumstances. Courts determine whether a work is transformative by considering whether it adds new expression or meaning to the original work, and whether it uses the original work for a different purpose.

Both derivative and transformative works can be protected under copyright law, provided they add significant creative elements to the original work. However, the line between derivative and transformative works can be blurry, and determining whether a work is derivative or transformative often requires a case-by-case analysis. In the context of AI, this distinction becomes even more important as AI tools often generate content by drawing upon or transforming existing works.

How does the ‘monkey selfie’ case relate to AI-generated content?

The U.S. Copyright Office finally settled that only humans are eligible to copyright things. Neither monkeys nor AI is eligible to own copyright. Only humans can.

This case has significant implications for AI-generated content. If a work is created by an AI, without human intervention, can that work be copyrighted? And if so, who owns the copyright? The ‘monkey selfie’ case makes it clear AI-generated works may not be eligible for copyright protection because they are not the product of a human mind. However, this is a rapidly evolving area of law, and different jurisdictions have different approaches. For example, some propose that AI-generated works may be protected under copyright law, with the copyright owned by the person who made the arrangements for the creation of the work.

What does the case of Richard Prince tell us about the threshold for creativity in copyright law?

In the context of AI, this case suggests that even minimal alterations to an AI-generated work could potentially make the work eligible for copyright protection. However, it’s important to note that technically speaking the transformation needs to add a new expression, meaning, or message to the original work. The changes would need to exhibit a degree of creativity and add a new layer of meaning or interpretation to the work. This could have significant implications for AI-generated content, particularly in fields like AI-generated art or music, where AI tools often create new works by transforming existing works.

How does the concept of ‘fair use’ apply to AI-generated content?

Fair use is a legal doctrine that allows for the unlicensed use of copyright-protected works in certain circumstances. These circumstances can include criticism, comment, news reporting, teaching, scholarship, and research. The concept of fair use is particularly relevant in the context of AI-generated content, as AI often relies on large datasets, which may include copyrighted material, for training purposes.

However, determining whether the use of copyrighted material by AI qualifies as fair use can be complex and depends on a case-by-case analysis. Courts typically consider four factors in a fair use analysis: the purpose and character of the use, the nature of the copyrighted work, the amount and substantiality of the portion used in relation to the copyrighted work as a whole, and the effect of the use upon the potential market for or value of the copyrighted work.

In the context of AI, these factors could be interpreted in various ways. For instance, if an AI uses a large portion of a copyrighted work to generate a new work, this could potentially weigh against a finding of fair use. However, if the AI’s use of the copyrighted work is transformative and does not harm the potential market for the original work, this could potentially weigh in favor of a finding of fair use.

What are the potential legal implications of using AI to generate content?

The use of AI to generate content raises a host of legal questions and potential implications. One of the primary legal issues revolves around copyright ownership. If an AI generates a work, who owns the copyright to that work? Is it the developer who created the AI? Or is it the user who provided the inputs and parameters for the AI to generate the content? Or does the work fall into the public domain because it lacks human authorship?

Another legal implication involves the potential for copyright infringement. If an AI is trained on copyrighted material and generates a work that is substantially similar to the copyrighted material, this could potentially constitute copyright infringement.

Furthermore, the use of AI to generate content could potentially raise issues related to privacy and data protection, particularly if the AI is trained on personal data or generates content that includes personal data.

What are the potential legal risks of using AI to generate content?

The use of AI to generate large-scale content can pose several potential legal risks. Here are some of the key risks expanded:

1. Copyright Infringement: One of the most significant risks is copyright infringement. AI models are trained on vast amounts of data, which often include copyrighted works. If the AI draws upon these copyrighted works to generate new content without obtaining the necessary permissions or licenses, this could constitute copyright infringement. This risk is particularly high when dealing with large-scale content, as the AI may inadvertently use substantial portions of copyrighted material in its output. For instance, an AI might generate a novel that closely resembles an existing copyrighted novel, or it might create a piece of music that includes significant elements of a copyrighted song. In such cases, the copyright owners could potentially sue for damages, leading to costly legal battles.
2. Unclear Ownership: Another major risk is the unclear ownership of AI-generated works. Copyright law traditionally protects works created by humans, and it’s not entirely clear how these laws apply to works created by AI. If an AI generates a work independently, without significant human intervention, it’s unclear who owns the copyright to that work. Is it the developer who created the AI? Is it the user who provided the inputs and parameters for the AI to generate the content? Or does the work fall into the public domain because it lacks human authorship? These are complex issues that are still being explored in legal circles, and the answers could have significant implications for the future of AI and creative work.
3. Fair Use Disputes: The doctrine of fair use allows for the use of copyrighted material without permission in certain circumstances, such as criticism, comment, news reporting, teaching, scholarship, and research. However, determining whether a use falls within the scope of fair use often requires a case-by-case analysis, taking into account factors such as the purpose and character of the use, the nature of the copyrighted work, the amount and substantiality of the portion used, and the effect of the use on the market for the original work. If the AI’s use of copyrighted material is claimed to be transformative and thus falls under fair use, this could lead to disputes over whether the use was indeed transformative, whether it meets the other criteria for fair use, and whether it infringes on the rights of the copyright owner.
4. Liability for User-Generated Content: If you’re operating a platform that allows users to generate content using AI, you could potentially face legal risks related to user-generated content. For instance, a user might use the AI to generate content that infringes on someone else’s copyright, defames someone, or violates other laws. Depending on the jurisdiction and the specific circumstances, you could potentially be held liable for this content, even if you didn’t directly create it.
5. Privacy and Data Protection Issues: AI models are often trained on large datasets that may include personal data. If you’re using AI to generate content, you need to ensure that you’re complying with all applicable privacy and data protection laws. This includes obtaining any necessary consents to use personal data, anonymizing data where possible, and implementing appropriate security measures to protect the data.

6. Defamation: AI systems, especially those generating large-scale content like novels or articles, could potentially make false statements that harm the reputation of individuals or companies. Even if unintended, such statements could lead to defamation lawsuits.
7. Right of Publicity Violations: The right of publicity is the right of an individual to control the commercial use of his or her name, image, likeness, or other aspects of their identity. If an AI uses someone’s name or likeness without consent, especially in a commercial context, it could lead to legal issues.
8. Ethical and Moral Rights Issues: Beyond legal risks, there are also ethical considerations. For instance, if an AI generates content that is offensive or harmful, it could lead to backlash and reputational damage. Additionally, in some jurisdictions, authors have moral rights in their works, such as the right to be recognized as the author of the work and the right to object to derogatory treatment of the work. It’s unclear how these rights apply to AI-generated content.
9. International Law Variations: Copyright laws vary significantly from country to country. What might be considered fair use in one country could be seen as infringement in another. If AI-generated content is distributed internationally, it could potentially run afoul of international copyright laws.
10. Patent Infringement: If the AI system is using patented algorithms or techniques without the necessary licenses, it could potentially lead to patent infringement claims.
11. Trade Secret Disclosure: If the AI system has been trained on proprietary data or uses proprietary algorithms, there could be a risk of unintentional trade secret disclosure.

What are some examples of AI-generated content that could potentially be copyrighted?

AI-generated content that could potentially be copyrighted includes any original work that exhibits a degree of creativity and is the result of human input or intervention. Here are some larger-scale examples:

1. AI-Generated Text: AI has the potential to generate extensive written content such as novels, comprehensive guides, full-length courses, or even entire encyclopedias akin to Wikipedia. If an AI, like ChatGPT, generates a unique arrangement of facts or information in these formats, that output could potentially be copyrightable. However, simply changing the order of facts in the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.
2. AI-Generated Art: AI can be used to create complex visual content such as feature-length animated films, documentaries, photo books, or large-scale digital art installations. If an AI generates a unique piece of art in these formats, that output could potentially be copyrightable. However, simply altering the color palette or composition of the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.
3. AI-Generated Music: AI has the capability to compose intricate musical pieces such as symphonies, full-length music albums, or scores for films and video games. If an AI generates a unique piece of music in these formats, that output could potentially be copyrightable. However, simply changing the tempo or key of the AI’s output may not be enough to claim copyright, as the changes would need to exhibit a degree of creativity.

In all these cases, it’s important to note that the bar for creativity is low, but not non-existent. The changes would need to exhibit a degree of creativity, such as a unique selection and arrangement of data or a unique expression of ideas.