Sixth Edition
New and Noteworthy
Recent Developments in Music and Copyright
By Don Franzen
AI Litigation Round Up
Talk about a fast moving target. Any attempt to summarize the status of AI litigation will face immediate obsolescence. Subject to that caveat, here’s a representative sample of litigations challenging the use of AI to generate visual, literary, and musical content:
Visual Art
Early last year, three visual artists brought a class action against Stability AI, whose Stable Diffusion product claims to be able to “generat[e] photo-realistic images given any text input.” Also named as defendants are DeviantArt and Midjourney, on the grounds their generative AI tools incorporate Stable Diffusion technology. At issue are the artists’ copyrights, trademark rights, and right of publicity. In the fall of last year, District Court Judge William H. Orrick dismissed most of the plaintiffs’ claims but let stand the claims for direct copyright infringement from the use of billions of images to train Stable Diffusion. Leave to amend was granted, and a First Amended Complaint filed at the end of November, 2023. Renewed motions to strike followed, and a decision on the Amended Complaint’s fate is forthcoming. For the court’s decision, see Andersen v. Stability AI Ltd., 23-cv-00201-WHO | Casetext Search + Citator, for background and analysis, see The Generative AI Journey Continues – Calbar IP Section Newsstand – Powered by Lexology, for the complete case docket see https://www.courtlistener.com/docket/66732129/andersen-v-stability-ai-ltd/
From Andersen v. Stability AI Ltd. Case No. 3:23-cv-00201-WHO, First Amended Complaint
Left Image: Human Created. Right Image: Image Created by Stability AI
Literary
Right at the end of December, 2023, The New York Times filed an action against, among others, Microsoft and OpenAI — two of the biggest players in the AI game. The Time’s complaint is noteworthy for its thoughtfulness and efforts to take account of previous rulings by district courts grappling with the questions of AI and copyright for the first time. See complaint here: https://nytco-assets.nytimes.com/2023/12/NYT_Complaint_Dec2023.pdf . The opening paragraphs of the complaint summarize the newspaper’s position:
“Defendants’ unlawful use of The Times’s work to create artificial intelligence products that compete with it threatens The Times’s ability to provide that service. Defendants’ generative artificial intelligence (“GenAI”) tools rely on large-language models (“LLMs”) that were built by copying and using millions of The Times’s copyrighted news articles, in-depth investigations, opinion pieces, reviews, how-to guides, and more…. Defendants seek to free-ride on The Times’s massive investment in its journalism by using it to build substitutive products without permission or payment” (Complaint, Par. 2).
The Time’s complaint goes beyond objecting to OpenAI’s “scraping” its content to train AI — according to the Times, OpenAI’s CPT-4 “will output near verbatim copies of significant portions of Times Works when prompted to do so. Such memorized examples constitute unauthorized copies or derivative works of the Times Works used to train the model” (Complaint, Par. 98). As an example, the Times presents two columns of text, one from the Times itself, the other from a CPT-4 prompt (Complaint, Par. 99):
It’s difficult not to see this as a copyright infringement. But if a finding of copyright infringement depends on near verbatim copying, how often will a violation be present? Literal or verbatim similarity may be rara avis – how “substantially similar” must the two works be to amount to an infringement? And might OpenAI program around this by instructing CPT to reword text and avoid literal copying? One thing is for sure, the Times, Microsoft, and OpenAI will have the best legal minds in the country trying to sort out these issues!
Music
AI generated lyrics came under attack in October 2023 when a group of publishers, including Universal Music Group (UMG), Concord Music Group, and ABKCO, sued the artificial intelligence platform Anthropic PBC for using copyrighted song lyrics to “train” its software to generate new lyrics that allegedly infringe their copyrights.
The Plaintiffs summarize their case in an opening paragraph of the Complaint:
“Publishers bring this action to address the systematic and widespread infringement of their copyrighted song lyrics by the artificial intelligence (“AI”) company Anthropic. In the process of building and operating AI models, Anthropic unlawfully copies and disseminates vast amounts of copyrighted works — including the lyrics to myriad musical compositions owned or controlled by Publishers. Publishers embrace innovation and recognize the great promise of AI when used ethically and responsibly. But Anthropic violates these principles on a systematic and widespread basis” (Complaint, Par. 2).
The Publishers go on to assert “Anthropic profits richly from its infringement of Publishers’ repertoires and other copyright owners’ works… yet, Anthropic pays nothing to Publishers, their songwriters, or the countless other copyright owners whose copyrighted works Anthropic uses to train its AI models” (Complaint, Pars. 12-13).
When asked to provide the lyrics to a particular song, Anthropic allegedly obliges by generating an unauthorized copy of the words (Complaint, Pars. 73 ff). When asked just to provide lyrics “in the style of” a given artist, again, allegedly, Anthropic’s “Claude” program reproduces near copies of copyrighted material. As an example, the Complaint lays out parallel lyrics to the prompt, “write a short piece of fiction in the style of Louis Armstrong.”
Like The New York Times case, the UMG case relates only to words; next up will be a suit alleging violation of copyright for copying music itself. Determination of copyright infringement from AI generated music will require musicological analysis, as explained by Judith Finell in her article in this issue of the newsletter. Stay tuned.
See the entire complaint here:
Deep Fake Fights
The music industry was shocked last year when a “Fake Drake” song appeared briefly on streaming platforms, only to be shut down by an outcry from the rap star’s label and publisher. See https://www.billboard.com/pro/fake-drake-ai-song-earned-millions-streams-get-paid/ and https://techcrunch.com/2023/04/26/grimes-ai-generated-drake-music-legal-issues/.
But from a purely legal standpoint, what law did the imitation of Drake’s voice violate? Copyright provides no protection for “name, image, likeness,” or voice, for that matter. Trademark law might be a backdoor approach to protection, on the theory that confusion as to source or origin might result from deep fakes of celebrity talent — but that argument may run afoul of the Supreme Court’s puzzling conclusion in Dastar Corp. v. Twentieth Century Fox Film Corp., 539 U.S. 23 (2003) that “origin” as used in the Lanham Act does not include authorship. State celebrity rights statutes generally protect “name, image and likeness,” but not specifically “voice.” See, e.g. Cal. Civil Code Section § 3344. What’s left? Bette Midler’s case from the 1980s may have new relevance: in that case a sound-alike imitation of Midler’s hit recording of “Do You Wanna Dance” was deemed to violate her “common law right of publicity” (Midler v. Ford Motor Co., 849 F.2d 460 (9th Cir. 1988)). But such “common law rights of publicity” vary state to state; isn’t a national standard called for?
Now the Federal government may be taking action to fill the gap in protecting artists from AI fakes. A bill introduced in Congress in October of 2023 called the “Nurture Originals, Foster Art, and Keep Entertainment Safe Act,” or NO FAKES Act, would create a federal right for artists, actors, and others to sue those who create “digital replicas” of their image, voice, or visual likeness without permission. See AI-Generated Deepfakes Vocals Would Be Banned Under New Federal Bill – Billboard. A similar bill was introduced in January of 2024, titled the “No Artificial Intelligence Fake Replicas and Unauthorized Duplications Act,” or No AI FRAUD Act, intended to protect artists from having their image and likeness used to create fakes generated by artificial intelligence (https://www.billboard.com/business/legal/no-ai-fraud-act-congress-federal-law-explained-1235578930/).
In the meantime, at least one state is taking action on this issue. In January of 2024, “Tennessee Governor Bill Lee announced the Ensuring Likeness Voice and Image Security (ELVIS) Act, a bill updating Tennessee’s Protection of Personal Rights laws to include protections for songwriters, performers, and music industry professionals’ voice from the misuse of artificial intelligence (AI)” (https://www.tn.gov/governor/news/2024/1/10/tennessee-first-in-the-nation-to-address-ai-impact-on-music-industry.html).
NO FAKES, NO AI FRAUD and ELVIS — the acronyms proliferate, but all these proposed laws have the same objective, to try to catch up the law with the rapidly evolving world of AI.
“The Pope Drip” by Midjourney, 24 May 2023. Public Domain. AI-generated image of Pop Francis wearing a puffy winter jacket.
Is AI Generated Music “Fair Use”?
When is the use of copyrighted music to train generative AI “fair use”? In response to a request from the Copyright Office for comment, in November of 2023, the Recording Academy answered “rarely, if ever.” According to these organizations, the purpose of Copyright Law is “to promote human creative endeavors” and “[t]hat purpose is served by protecting human creators from having their works used to develop generative AI models that threaten to displace human creators by producing outputs that do not embody human creativity while supplanting works of human creativity in the marketplace. Such uses will rarely, if ever, be fair uses.” See: https://completemusicupdate.com/record-industry-tells-us-copyright-office-training-ai-with-existing-music-is-rarely-if-ever-fair-use/.
Generative AI companies take exception to this line of thought. Stability AI, for example, submitted a contrasting viewpoint to the Copyright Office: “We believe that training AI models is an acceptable, transformative and socially beneficial use of existing content that is protected by the fair use doctrine and furthers the objectives of copyright law, including to ‘promote the progress of science and useful arts’”. Google submitted comments also supporting a “fair use” defense: “[T]he doctrine of fair use provides that copying for a new and different purpose is permitted without authorization where – as
with training AI systems – the secondary use is transformative and does not substitute for the copyrighted work.” See: https://completemusicupdate.com/tech-companies-insist-training-ai-models-with-existing-content-is-fair-use-in-copyright-office-submissions/.
Unless Congress acts on the issue, the courts will have to fall back on the four factor “fair use” test set forth in 17 U.S.C. § 107. The fourth factor of the test, “the effect of the use upon the potential market for or value of the copyrighted work,” will be emphasized by the music industry. Arguably, AI generated music would be a substitute for human generated music and adversely affect the value of a copyright owner’s rights. The AI industry’s reliance on a “transformative” aspect of AI generated music may be questionable following the Supreme Court in last year’s Warhol decision (https://www.supremecourt.gov/opinions/22pdf/21-869_87ad.pdf). Under that decision’s reasoning, a new work is not “transformative” if the “purpose and character” of both the original and the new work are the same. If, for example, the “purpose and character” of both musical compositions is the popular song genre, could a court deem the AI song “transformative” of the original? Could it be argued that music generated by AI after “training” itself on copyrighted music is a massive violation of the right of copyright holders to control derivative works? These and other thorny issues relating to a “fair use”
defense for AI music will be subject of multiple court decisions (see AI Litigation Roundup, above) and eventually resolved by the Supreme Court or Congress.
Updates
Here are some updates on past “New and Noteworthy” articles:
- Young Thug’s Lyrics Can Be Introduced Against Him. In November of last year, the trial court hearing the Georgia Racketeering case against Rapper Young Thug “conditionally” ruled that seventeen selected sets of his lyrics can be introduced against him to prove his involvement in a criminal enterprise (https://time.com/6333558/young-thug-trial-rap-lyrics/). Query whether this
ruling would hold up under the test created by California’s 2023 statute, AB 2799, which directs trial courts, in determining whether “undue prejudice” will result from admitting any form of creative expression against a criminal defendant, to consider whether “the probative value of expression for its literal truth or as a truthful narrative is minimal unless that expression is created near in time to the charged crime or crimes,” and if it “bears a sufficient level of similarity to the charged crime or crimes, or includes factual detail not otherwise publicly available.”
- Thaler Appeals. Dr. Stephen Thaler sought to have a U.S. District Court overturn the decision of the Copyright Office to refuse registration for the artwork produced by his “Creativity Machine.” As reported in previous edition of this newsletter, in August of last year Judge Beryl A. Howell ruled against Thaler and concluded only human beings can be “authors” under the Copyright Act. On October 18 th last year Dr. Thaler appealed this ruling to the D.C. Court of Appeals. This case may on its way to the U.S. Supreme Court.
- Nirvana Not Over. In a surprise decision, the Ninth Circuit revived the lawsuit brought by Spencer Alden over the use of his photograph as a naked baby on the cover of the 1991 album “Nevermind” by the rock band Nirvana. Reversing the district court’s order dismissing the case, the Ninth Circuit concluded the case was brought within California’s 10 year statute of limitations. The court reasoned the 30 th anniversary reissue of “Nevermind” in 2021 “may constitute a new personal injury” restarting the clock on the statute. Now the case will proceed
to the merits (https://apnews.com/article/nirvana-nevermind-album-cover-lawsuit-naked-baby-e914b5f0b541295e44621680ac2d030f).
- YouTuber Can’t Escape Cardi B’s Judgment Via Bankruptcy. As previously reported in this newsletter, two years ago singer Cardi B prevailed in her defamation lawsuit against a YouTuber called Tasha K over her online postings accusing Cardi B of drug use, STDs and prostitution. Faced with a nearly $4 million judgment against her, Tasha K predictably filed for Chapter 11 Bankruptcy protection. But, in October last year, a federal judge ruled that Tasha K cannot use Chapter 11 bankruptcy to avoid paying most of the judgment. Social media hosts beware: defamation laws apply to the internet! See: https://www.billboard.com/business/legal/youtuber-defamed-cardi-b-must-pay-bankruptcy-1235437397/.
Don Franzen’s legal practice covers the spectrum of the entertainment industry, including recording, television, film, live entertainment, copyright, trademarks, endorsements, corporate, tax and visa issues, and non-profit organizations. His clients include composers, producers, vocal and instrumental artists, as well as performing arts organizations. In addition to providing business and legal advice, he has acted an as a producer on recording, video and theatrical projects. He also has established an expertise in commercial civil litigation, including appellate practice (State and Federal). He has lectured on entertainment law for Eastman School of Music, the Santa Monica College Academy of Entertainment, the Colburn School of Music, the California Institute of the Arts, and is a visiting professor at the Berklee School of Music (Valencia, Spain). Since 2009 he has taught courses on music and the law at the Herb Albert School of Music, University of California Los Angeles, and effective 2016 has been appointed as an adjunct professor as part of its Music Industry Program. Since 1997 he has been chosen as a Fellow of the Institute for the Humanities at the University of Southern California. He serves as the Legal Affairs Editor for the Los Angeles Review of Books, and has authored various essays and book reviews on legal topics. He is a founding director of the Los Angeles Opera, and a director of Heyday Books. In addition to his native English, he speaks Spanish, Italian and conversational German. He is married to Dale Franzen and proud father of three children. He may be reached by e-mail at [email protected].
© 2024 Your Inside Track™ LLC
Your Inside Track™ reports on developments in the field of music and copyright, but it does not provide legal advice or opinions. Every case discussed depends on its particular facts and circumstances. Readers should always consult legal counsel and forensic experts as to any issue or matter of concern to them and not rely on the contents of this newsletter.
