Artificial intelligence is causing a stir among the arts community as concerns surrounding the future of artistic occupations loom. Some wonder if their work will become irrelevant, and will the work they produce be protected?
We asked Nazareth Pantaloni III, copyright program librarian at IU Libraries, to answer several burning questions about AI in the context of the arts, intellectual property and copyright law.
Question: Currently, who gets credit when someone writes a song or creates a piece of art using AI?
Answer: AI raises profound questions about the nature of human creativity. But issues of artistic credit were already complicated before the advent of AI, involving such factors as stylistic innovation, inspiration or customary practices in different creative industries.
For example, many contemporary artists use art fabricators to execute their ideas and to produce works of art for which the brand-name artist, not their assistants, gets credit. In the music industry, popular performers are often credited as the writers of the songs they sing, but did not necessarily help to compose, so they can collect a share of the performance royalties.
As a legal matter, copyright law establishes what creative expressions are copyrightable and who owns the copyright in copyrighted works, not who gets credit for them. Film credits often list the names of dozens of talented people who contributed to the creation of a film, but typically just one copyright owner — namely, the film company that produced it. Ownership of the copyright in a work belongs to the creator or creators of the work, unless they have produced it as a work for hire, as an employee, in which case the employer or person for whom the work was created will usually be considered the owner of any copyright in the work.
More significant to AI-generated works is the issue of copyrightability, or what copyright protects. Copyright only protects original works by human creators. Under U.S. copyright law, copyright protection does not extend to non-human creators, which is an obvious problem for AI-generated works. The Copyright Office will not register works produced by a machine or mechanical process unless there is some copyrightable input or intervention by a human creator.
Two cases from the past year illustrate the practical implications of the legal requirement that a work must be authored, at least in part, by a human creator. In a case that is currently being litigated in a federal trial court, Thaler v. Perlmutter, the inventor of an AI image generator sued the U.S. Copyright Office for its refusal to register a work created exclusively by AI. The federal District Court for the District of Columbia has not yet announced their decision, but it is likely that they will uphold the longstanding legal principle that only works by human creators are copyrightable.
In October 2022, the U.S. Copyright Office canceled a copyright registration for a graphic novel — “Zarya of the Dawn” by Kris Kashtanova — that was produced using the AI-generative tool Midjourney, when it was publicized that the AI-generated work had been successfully registered with the Copyright Office. In February 2023, the Copyright Office granted a new copyright registration for the graphic novel’s text, which was authored by Kashtanova, as well as the author’s selection, coordination and arrangement of their text with the AI-generated visual elements. So, while Kashtanova is credited as the creator of the graphic novel, copyright protection only extends to the human-authored parts of the work, not to those visual elements produced by the text-to-image generator Midjourney.
On March 16, 2023, the Copyright Office issued guidance for works containing material generated by AI, reiterating the human authorship requirement, but allowing that it is possible for a work that contains AI-generated material to include sufficient human authorship to support a copyright registration when a creator selects, arranges or modifies AI-generated material in a sufficiently original way. The Copyright Office emphasized that copyright only protects the human-authored parts of a work under U.S. copyright law, as well as their current rules for reviewing applications for copyright registration.
Q: What is happening in terms of updating intellectual property law in the U.S. (and internationally) and the rules and restrictions around using someone else’s intellectual property now that AI is widely accessible?
A: At present, countries are relying on their current laws and regulations to navigate the intellectual property issues raised by the use of massive collections of creative works as data for machine learning or the training of generative AI systems.
In the European Union, for example, the Copyright in the Digital Single Market Directive adopted in 2019 allows for text and data mining of works lawfully accessible on the public internet for research purposes by academic and cultural heritage research institutions. For commercial purposes, such as the use of data scraped from the internet to train AI systems, the CDSM Directive permits copyright owners and creators to opt out of the use of their works for text and data mining, and thereby reserve the right to negotiate fees for the licensed use of their copyrighted works for such purposes.
In the United States and United Kingdom, where answers to the legal questions raised by AI are less clear under current law, the issues are being litigated in at least two cases in the U.S. and a related lawsuit in the U.K. All three cases name Stability AI, the developer of the text-to-image generator Stable Diffusion, as a defendant.
The plaintiffs include artists and illustrators, along with Getty Images, who are rights holders. They argue that AI generators use their work without permission to compile data sets for training a platform’s algorithms. The AI image generators can then be used to produce images that are in the style of — or substantially similar to the work of — a particular creator, and therefore infringe the right of artists to control the creation of derivative works, defined as a work based on a preexisting work or works that have been adapted, recast or transformed.
The defendant developers of AI platforms are generally claiming that their use of images and other copyrighted works scraped from the internet is a transformative fair use that uses the works for a different purpose from the originals, as data to train AI algorithms, in a way that does not impact the potential market or value of the original works.
AI presents a complex array of legal issues that will take years to resolve. In the meantime, the U.S. Copyright Office is hosting public listening sessions with interested parties to consider the legal and policy issues implicated by AI. They expect to announce a call for public comments on copyright and AI by the end of this year.
Q: Writers, musicians and artists fear their occupations may be negatively impacted by AI and are calling for regulations to protect their work (i.e. Hollywood writers strike). How much of these regulations will be left up to corporations and organizations to create, and what do you foresee being federally regulated?
A: Decisions from federal courts will, hopefully, give creators and AI developers a clearer sense of their legal rights and responsibilities. That could prompt companies to negotiate royalty agreements for the use of copyrighted works used by AI platforms, or to proceed as they currently are.
Some industry and legal observers are calling for a so-called “Spotify solution,” which would involve the creation of collective licensing mechanisms to resolve some of AI’s legal questions, just as the advent of legal music streaming platforms put an end to the recording industry’s lawsuits for file-sharing.
Courts can only decide individual cases, so they have a limited ability to impose an industry-wide solution, while Congress is often unwilling to intervene too hastily in a new industry, for fear that they might disrupt it. If the recent history of technological innovation is any indication, Congress will be looking to interested parties to begin to resolve the issues, and then possibly step in at some point with legislation to clarify or extend copyright law in this area or to establish a means for licensing works for use by AI platforms.
Q: Do you have any advice for artists to protect their intellectual property as this wave of AI begins to roll?
A: The history of copyright law is a function of technological change, from the printing press to photography, sound recordings, radio, movies, television and the internet. It is not clear how copyright law will adapt to AI, but the technology is clearly going to transform the ways in which people create and consume art and music.
While copyright is automatic from the moment an original work of human authorship is expressed in a tangible way, the most reliable way to protect a copyrighted work in this time of uncertainty is to register it with the U.S. Copyright Office within three months of making it available to the general public. Copyright registration establishes the right to certain types of damages and attorneys’ fees and is a requirement for copyright infringement claims.