IPR In Licensing AI-Generated Lyrics.
1. Understanding IPR in AI-Generated Lyrics
AI-generated lyrics are textual outputs created by AI models, trained on large datasets of music and lyrics. The key IPR concerns are:
Copyright Ownership:
Traditionally, copyright protects original works created by a human author. The big question is whether AI-generated content qualifies for copyright protection.
Licensing:
Licensing involves granting someone permission to use the work under certain conditions. AI-generated works complicate licensing because ownership may be ambiguous.
Derivative Works:
Lyrics generated by AI may be inspired by pre-existing songs. Using AI to create new lyrics could trigger infringement claims if the AI “copies” too closely from copyrighted works.
2. Legal Issues in AI-Generated Lyrics Licensing
Authorship: Most jurisdictions currently require a human author for copyright. AI alone cannot be considered an author.
Ownership: If a human provides prompts and selects outputs, the human may hold the copyright.
Infringement Risks: AI can generate lyrics that unintentionally resemble copyrighted works.
Licensing Models: Contracts must clearly define rights—who can use, distribute, modify, or commercialize AI-generated lyrics.
3. Key Case Laws and Examples
I’ll detail five important cases related to AI, copyright, and music, and explain how they relate to AI-generated lyrics:
Case 1: Naruto v. Slater (Monkey Selfie Case, 2018, US)
Facts:
A monkey took a selfie using a wildlife photographer’s camera. PETA tried to claim copyright on behalf of the monkey.
Holding:
The US court ruled that non-human entities cannot hold copyright.
Relevance to AI lyrics:
This case is often cited to argue that AI alone cannot be the author of copyrighted material.
In licensing AI-generated lyrics, this means that copyright must be assigned to a human who contributed creatively to the work.
Case 2: Feist Publications, Inc. v. Rural Telephone Service Co. (1991, US)
Facts:
Feist published a phone directory, and Rural sued for copying its white pages.
Holding:
Facts are not copyrightable unless they have original creative expression.
Relevance:
AI-generated lyrics must demonstrate originality, not just be a compilation of existing text.
If AI outputs lyrics that are purely derivative of existing songs, they may not qualify for copyright protection.
Case 3: Naruto v. Slater Extended to AI: Thaler v. Commissioner of Patents (DABUS Case, 2021, Australia)
Facts:
Dr. Stephen Thaler applied for a patent listing an AI (DABUS) as the inventor.
Holding:
Australian courts allowed AI-generated inventions to be recognized in some circumstances, but US and UK courts rejected AI as an inventor.
Relevance:
While this is patent law, it shows the global debate on whether non-human creators can hold IP rights.
For licensing AI-generated lyrics, it strengthens the argument that the human operator must hold copyright for the AI-generated work to be licensable.
Case 4: Warner Music Group v. Resso (Hypothetical AI Lyrics Infringement, 2023, Brazil)
Facts:
Resso, an AI music platform, generated lyrics similar to songs owned by Warner Music.
Holding:
The Brazilian court emphasized that AI-generated content can infringe copyright if it substantially reproduces existing works.
Relevance:
Licensing agreements must warrant originality.
Platforms distributing AI-generated lyrics need clear indemnity clauses to protect against infringement claims.
Case 5: Authors Guild v. Google, Inc. (2015, US)
Facts:
Google scanned millions of books and made snippets searchable. Authors sued for copyright infringement.
Holding:
The court ruled Google’s use as fair use, focusing on transformative use rather than full reproduction.
Relevance:
AI-generated lyrics often train on existing songs. Licensing must consider transformative vs derivative use.
This affects royalty structures and permissions for AI-generated works.
Case 6: GitHub Copilot Licensing Debate (OpenAI / Microsoft AI tools, 2022-2023)
Facts:
Copilot suggested code snippets similar to open-source repositories. Developers sued claiming copyright infringement.
Relevance:
Similar to AI lyrics generation, if AI replicates copyrighted lyrics, licensors may face legal exposure.
Highlights the importance of data source transparency when licensing AI outputs.
4. Practical Licensing Approach for AI-Generated Lyrics
Clarify Ownership:
Determine the human author or company controlling the AI.
Explicitly assign copyright in contracts.
Originality Warranty:
Include clauses confirming the lyrics are original and non-infringing.
Royalty and Usage Terms:
Decide if the license is exclusive or non-exclusive.
Specify if the license covers derivative works, public performance, and digital distribution.
Indemnification Clause:
The licensor should agree to protect the licensee against copyright claims.
Transparency of AI Tools:
Document training data sources and human involvement to defend copyright claims.
✅ Key Takeaways
AI cannot currently hold copyright independently; human input is crucial.
Originality and non-infringement are central for licensing AI-generated lyrics.
Licensing agreements must carefully define ownership, usage rights, and liability.
Case law is evolving, but current trends favor human authorship and caution against derivative outputs.
RELATED Blog
comments