Introduction
The evolving legal landscape surrounding the use of copyrighted content by AI companies is marked by significant court rulings and ongoing litigation. These cases primarily focus on whether the use of copyrighted material to train AI models constitutes fair use under US copyright law. The outcomes of these legal battles have profound implications for the future of AI technology and the protection of intellectual property rights.
Description
AI companies assert that their use of copyrighted content [3], such as music and lyrics [3], qualifies as fair use [3] [5], particularly emphasizing the transformative nature of their applications [3]. A significant ruling from the US District Court for the Northern District of California determined that using copyrighted works to train large language models (LLMs) constitutes fair use under US copyright law [6]. This decision arose from the case Bartz v [6]. Anthropic PBC [6] [7], where authors claimed their books were used without permission to train Anthropic’s Claude AI system [6]. The court assessed various factors of fair use [6], including the purpose of the use [6], the nature of the copyrighted work [6], the amount used [5] [6], and the impact on the market for the original works [6]. It concluded that training LLMs is fair use [6], as the models did not replicate the authors’ works but rather utilized them to learn language patterns [6]. However, the case also revealed that Anthropic had allegedly downloaded up to 7 million books from pirate websites [7], leading to claims of significant copyright infringement. The court noted that while Anthropic’s AI training could constitute fair use [7], the company violated copyright by maintaining pirated copies of the books in a central repository [7]. This aspect of the case is set to proceed to trial [6], with the potential for billions in damages.
In light of recent legal actions initiated by several major record labels against AI music generators Suno and Udio, the focus has shifted to the protection of sound recordings produced by these AI tools. The labels allege that both companies have infringed copyright by generating outputs that closely resemble existing sound recordings, seeking damages of $150,000 for each allegedly infringed track [5]. Udio is specifically accused of producing outputs with “striking resemblances” to well-known songs like “Dancing Queen” and “All I Want For Christmas Is You,” while Suno faces similar allegations regarding songs such as “I Got You (I Feel Good)” and “Johnny B. Goode.” Both companies utilize copyrighted music to train their models [4], generating outputs that raise significant legal questions about fair use under US copyright law. Suno [2] [3] [4] [5], in particular, is facing multiple lawsuits alleging that it unlawfully trained its AI model using copyrighted material [2].
Suno and Udio argue that their technology is transformative [5], as it synthesizes new [5], original outputs rather than merely copying existing songs [5]. Central to the litigation is whether the generative AI music is sufficiently transformative [5], meaning it adds new meaning [5], expression [5], or value to the original works [5]. The court will assess the amount and substantiality of the portions of songs that were copied [5], considering both qualitative and quantitative factors [5]. Additionally, the impact of the generative AI outputs on the market value of the original recordings will be evaluated [5], particularly whether the AI-generated music substitutes for the original songs [5]. The effect of such content on the demand for original works is a critical factor in fair use analysis [3], and AI-generated music could compete directly with human-made music [3], potentially favoring plaintiffs in market effect arguments [3].
Plaintiffs may leverage insights from the Anthropic case to challenge AI companies’ fair use defenses [3]. Record labels bear the responsibility of safeguarding their artists’ rights [4], and if AI start-ups maintain that their outputs do not infringe on copyright [4], labels may need to consider extending personality rights to encompass sound [4]. To strengthen their cases [3], plaintiffs should focus on demonstrating market harm due to AI-generated content diluting demand for original works [3], rather than merely claiming lost licensing opportunities [3]. Evidence of market dilution or substitution will be crucial [3], as will documentation of prior licensing attempts to AI companies [3], which can substantiate claims of lost revenue [3]. Collaboration between AI start-ups and artists is essential to create a licensing framework that respects the cultural significance of music [4].
Recent judicial rulings suggest that licensing agreements can mitigate many challenges for generative AI companies [3]. If copyright material is obtained legally [3], either through licensing or purchase [3], it is likely to be viewed as transformative use [3]. Early licensing negotiations may also help copyright holders avoid litigation [3], as the market for licensing content to AI companies continues to grow [3], exemplified by recent agreements like The New York Times licensing its content to Amazon for LLM training [3]. While AI holds the potential to foster innovation in music creation [4], its deployment must adhere to copyright laws to protect the integrity of the music industry [4]. The outcomes of the Suno and Udio cases may depend on judicial interpretations of market effects [3], with both companies arguing for fair use based on transformative use [3], but the complexities of music copyright law may complicate plaintiffs’ infringement claims [3]. Unlike literary works [1] [3], musical recordings involve multiple copyrightable elements [3], which may weaken the labels’ positions [3].
Additionally, many creators are pursuing lawsuits against generative AI companies [1], alleging large-scale copyright infringement due to the use of their works for training AI models [1]. Notable cases include Andersen v [1]. Stability AI [1], Zhang v [1]. Google [1] [4], and Tremblay v [1]. OpenAI [1], all concerning the use of artistic and literary works to train diffusion models and LLMs [1]. The legal landscape is evolving [1] [6], with issues surrounding piracy [1], web scraping [1], and the copyrightability of generated artwork being central to these disputes [1]. Ongoing litigation in multiple jurisdictions [1], including Massachusetts and New York [1], highlights the complexities and challenges faced by both creators and AI companies in navigating copyright law. Legal experts are closely monitoring these developments, as they hold significant implications for the future of creative work and copyright issues related to AI technologies.
Conclusion
The ongoing legal disputes between AI companies and copyright holders underscore the complexities of applying traditional copyright laws to emerging technologies. The outcomes of these cases will significantly impact the development and deployment of AI technologies, particularly in creative industries. As the legal framework continues to evolve, collaboration between AI developers and content creators will be crucial in establishing a balanced approach that fosters innovation while protecting intellectual property rights.
References
[1] https://aiwatch.dog/lawsuits
[2] https://www.wbur.org/news/2025/07/23/suno-ai-music-lawsuit-copyright-ip
[3] https://www.jdsupra.com/legalnews/harmonizing-ai-and-copyright-fair-use-3844953/
[4] https://www.thegazellenews.com/harmony-or-infringement-legal-industry-implications-of-ai-generated-music-in-age-of-copyright/
[5] https://www.thetimes.com.au/world/32272-record-labels-are-suing-tech-companies-for-copying-classic-songs-%E2%80%93-and-the-results-could-shape-the-legal-future-of-generative-ai
[6] https://www.legallyblackip.com/blog/zrvmkqhy0kmef5bajfngeq0f4w0oy3
[7] https://www.bnnbloomberg.ca/business/artificial-intelligence/2025/07/17/us-authors-suing-anthropic-can-band-together-in-copyright-class-action-judge-rules/
