Anthropic vs. UMG: Should Universal Have Focused on Fewer Songs?

The ongoing legal dispute between Anthropic and Universal Music Group (UMG) has sparked significant discussion about how copyright infringement claims should be handled in the age of AI. This blog draws insights from a TechTarget article by Esther Shittu, published on March 28, 2025 (Shittu, 2025), and an article by Christopher Wieduwilt, the AI Musicpreneur (Wieduwilt, 2025), both of which provide valuable perspectives on this case. 

A central question emerges: Would UMG have been more effective if they had focused on just two or three songs instead of presenting evidence for 500 songs? 

The judge in the case deemed UMG’s claim "too broad," (Shittu, para. 2, 2025) which may have weakened their argument. UMG and other publishers alleged that Anthropic's AI, Claude, reproduced exact copies of lyrics from major hits like "Uptown Funk" by Bruno Mars and "Halo" by Beyoncé (Shittu, para. 4, 2025).  However, a more precise and forensic approach might have yielded stronger results. 

Could a Forensic Approach Have Made the Difference?
Based on the judge’s comments (see next paragraph), it appears that UMG and the publishers may not have taken a detailed or what I would call out as a forensic approach to presenting their evidence:

"Although publishers have tried to clarify the scope of the proposed injunction, the details remain elusive and poorly defined...[because the complaint incorporated and referenced a 500 song list, with the judge referring to that song list as 'illustrative and non-exhaustive.'"  (Shittu, para. 3, 2025)

But, while Judge Eumi K. Lee denied UMG’s request for a preliminary injunction against Anthropic, ruling that publishers failed to demonstrate “irreparable harm, "the court granted publishers favorable discovery terms, allowing them to gather more evidence from Anthropic’s Claude AI system.  (Wieduwilt, para. 8, 2025)

What a Flawless Forensic Approach Might Look Like
From my experience in information management and data and records preservation, I believe that a forensic approach would have been nearly flawless if UMG had taken a detailed approach to presenting their evidence. Here’s what such an approach would entail:

1. Using a New, Untouched Computer - A brand-new computer, never connected to the internet or used for other purposes, would ensure the integrity of the evidence. This would prevent external interference or tampering; contamination of system metadata; questions about pre-existing files or activities on the device.
   
   Starting with a clean slate is critical to maintaining the chain of custody and ensuring that every action, from input to output, is traceable and uncontaminated.
   
   
2. Documenting Every Step - Publishers could have bolstered their case by meticulously recording the entire process. Video evidence showing the prompts entered into Claude, capturing the AI’s output in real time to demonstrate how it allegedly reproduced copyrighted lyrics. Highlighting specific matches between the AI-generated text and the original lyrics.
   Such documentation would leave no room for ambiguity and provide a clear, step-by-step trail of evidence.
   
   
3. Protecting Chain of Custody - Ensuring the integrity of the evidence involves. Timestamping every interaction with the AI system. Preserving system metadata to verify the authenticity of the outputs. Isolating the evidence from external threats—e.g., storing it in a secure, offline environment. These steps would make it difficult for any party to question the legitimacy of the evidence.
   
   
4. Narrowing the Scope - presenting evidence for 500 songs likely overwhelmed the court and diluted the strength of UMG’s argument. Instead, they could have focused on just two or three high-profile songs with undeniable similarities. 
   
   According to the TechTarget article, quoting Michael Bennett, the associate vice chancellor for data science and AI strategy at the University of Illinois in Chicago, the judge emphasized this point: 
   
   "If you’re trying to convince the court to give you injunctive relief against [a tech company] because you’re convinced that they used your copyrighted material to train their models, then your request ought to be as specific as possible. If it’s vague, then it’s going to be denied."  (Shittu, para. 9, 2025)
   
   A targeted approach would have provided the court with a clearer and more manageable case.

Lessons for the Music Industry

This case highlights key lessons for musicians and publishers navigating the intersection of AI and copyright law:

1. Specificity is Key: Broad claims are less effective than focused, detailed arguments.
2. Forensic Documentation: Every step of the process must be recorded and preserved to prevent ambiguity.
3. Chain of Custody: Using a new, dedicated computer to document evidence ensures integrity and avoids contamination.
4. Precision Over Volume: Presenting fewer, well-documented examples is more convincing than overwhelming the court with excessive evidence.

As a musician, I understand the frustration of seeing creative works potentially misused. However, as someone with a background in data preservation, I also recognize the importance of building a case that can withstand legal scrutiny. UMG’s decision to address 500 songs at once may have diluted their argument and made it harder for the court to evaluate their claim. As Christopher Wieduwilt insightfully observes, these disputes are a sign of how AI is reshaping copyright law and the music industry, underscoring the urgent need for new strategies to navigate these challenges (Wieduwilt, para. 8, 2025).

Final Takeaways
The Anthropic vs. UMG case is a powerful reminder of the challenges posed by AI to copyright law. For future cases, specificity, precision, and a flawless forensic approach will be critical for success.

References:

Shittu, E. (March 28, 2025), Anthropic's early lawsuit win pushes courts forward on fair use, TechTarget, https://www.techtarget.com/searchenterpriseai/news/366621658/Anthropics-early-lawsuit-win-pushes-courts-forward-on-fair-use#:~:text=On%20March%2025%2C%20a%20California,while%20the%20lawsuit%20was%20ongoing.

Wieduwilt, C. (March 26, 2025), Breaking: Judge Lee denies UMG’s request to block AI company Anthropic from using song lyrics for their AI chatbot Claude – Here’s what you need to know, The AI Musicpreneur, https://www.aimusicpreneur.com/ai-music-news/universal-music-anthropic-lawsuit-major-blow-judge-lee/