Ipr In Licensing AI-Generated Ar Assets.
IPR IN LICENSING AI-GENERATED ART ASSETS
1. Introduction
AI-generated art assets refer to creative works (such as images, music, video, 3D models, and more) produced with the help of artificial intelligence systems. AI algorithms, often trained on vast datasets, can autonomously create content that mimics or innovates upon existing styles, patterns, and artistic genres.
However, the licensing of AI-generated art poses unique legal challenges:
Who owns the AI-generated content?
How do traditional copyright and licensing models apply?
How should companies protect or monetize AI-generated content?
The legal ownership of AI-generated assets and their subsequent licensing are critical in industries such as gaming, advertising, film production, and digital art.
2. Types of AI-Generated Art Assets and Licensing Challenges
Digital Art (images, illustrations, paintings)
AI tools such as DALL·E or MidJourney produce digital art from text prompts.
Music (AI-composed songs, background scores)
Tools like AIVA or Amper Music generate musical compositions.
3D Models & Virtual Objects (avatars, virtual environments, game assets)
AI-driven systems generate 3D models for virtual worlds or video games.
Text-Based Content (poems, stories, scripts)
AI like GPT-3 generates written content based on prompts.
Licensing models for these AI assets often need to balance:
The role of the creator: Is the AI the "author," or does the user (prompt creator) own it?
The ownership of training data: AI learns from large datasets—who owns the data?
Rights of modification and distribution: Can others modify and resell AI-generated assets?
3. IPR Licensing Models for AI-Generated Art
(a) Exclusive Licensing
The licensee receives exclusive rights to use, modify, and distribute the AI-generated asset.
Common in industries where the asset (e.g., AI-generated music for a movie) needs full ownership rights.
(b) Non-Exclusive Licensing
The licensor retains the right to license the AI-generated asset to others while granting rights to the licensee.
Often used in the gaming industry, where assets like AI-generated skins or environments can be sold to multiple buyers.
(c) Royalty-Based Licensing
Licensees can sell or use the AI asset commercially, but a percentage of profits (royalties) must be paid to the original creator or AI platform.
Common in NFT licensing where the creator of AI-generated NFTs receives a percentage from secondary sales.
(d) Open Licensing
Licensor allows others to use, modify, or distribute AI-generated assets under open-source terms.
Popular for community-driven projects in virtual worlds, open-source AI art generators, or collaborative art platforms.
CASE LAWS (DETAILED ANALYSIS)
Case 1: Feist Publications, Inc. v. Rural Telephone Service Co. (1991, US)
Facts:
Rural Telephone published a telephone directory.
Feist copied parts of the directory and published a competing one. Rural claimed copyright infringement over the selection and arrangement of the data.
Judgment:
The Court ruled that facts themselves are not copyrightable, only creative expressions are.
Relevance to AI-Generated Art:
AI-generated content that replicates or uses publicly available data may not be automatically eligible for copyright unless the AI output involves originality (e.g., new artistic elements, creative interpretation).
The originality of the user’s input or prompt can influence copyright ownership in AI-generated assets.
Case 2: Naruto v. Slater (2018, US)
Facts:
A macaque monkey, Naruto, took a selfie with a photographer's camera.
The photographer filed a copyright claim, but the monkey was the "creator" of the image.
Legal Issue:
Can non-human creators hold copyright?
Judgment:
The court ruled that only humans can be considered as authors under US copyright law, so the monkey could not hold the copyright to the selfie.
Relevance:
AI systems generating art could present a similar challenge: Who owns the copyright for an AI-created asset?
If the AI is not seen as the "author," the person controlling or inputting prompts may be considered the author, meaning they can license the asset.
The owner of the AI system may also have a claim if the AI’s creativity is highly autonomous.
Case 3: Thales Visionix, Inc. v. United States (2017, US)
Facts:
Thales Visionix filed a patent for a motion-tracking system.
The U.S. government challenged the patent based on abstractness.
Judgment:
The court upheld the patent because the system involved technical improvement.
Relevance:
The case highlights the importance of demonstrating technical innovation in AI systems when applying for patents related to AI-driven asset generation tools.
Licensing of AI technology or algorithms that generate creative assets (e.g., AI art generation software) should focus on the technical uniqueness of the AI system or algorithm.
Case 4: Warhol Foundation v. Lynn Goldsmith (2019, US)
Facts:
The Warhol Foundation used a photograph by Lynn Goldsmith as the basis for a series of Andy Warhol’s artwork.
Goldsmith claimed copyright infringement.
Judgment:
The court ruled in favor of Goldsmith, stating that Warhol's adaptation was not transformative enough to outweigh the original copyright.
Relevance:
This case reinforces that transformative use—a key argument in fair use claims—will be scrutinized carefully in AI-generated content.
If an AI creates a piece that significantly alters or adds new creativity to an existing work, fair use may be invoked.
But if the work is seen as derivative, the original owner may retain licensing rights.
Case 5: Google Inc. v. Oracle America, Inc. (2021, US)
Facts:
Oracle sued Google, claiming that Google used Java code in its Android operating system without permission.
Legal Issue:
Whether Google's use of Java was fair use under copyright law.
Judgment:
The Court ruled that Google’s use of Java was fair use because it was transformative and involved new software development.
Relevance:
This case can be important for licensing AI-driven art, where the AI model uses existing works as training data.
If the AI is merely using existing data to generate new creations, it could argue fair use or transformative use.
The issue of data ownership for AI training datasets can influence licensing terms.
Case 6: Nike, Inc. v. StockX, Inc. (2021, US)
Facts:
StockX sold NFTs that represented limited edition Nike sneakers.
Nike claimed infringement, arguing that StockX misused its trademark and product designs in the digital space.
Judgment:
Nike won the case, asserting its exclusive trademark rights over the Nike brand and its iconic sneakers.
Relevance:
Trademark infringement and brand protection apply to AI-generated assets in the Metaverse.
AI-generated assets resembling branded items (like virtual sneakers) can lead to trademark disputes.
Licensing agreements for AI-generated assets in virtual worlds or games should address trademark concerns.
Case 7: Hermès International v. Rothschild (2022, US)
Facts:
Rothschild created “MetaBirkin” NFTs, which resembled Hermès' luxury Birkin bags.
Hermès sued for trademark infringement.
Judgment:
The court ruled in favor of Hermès, holding that NFTs created using trademarks without permission can lead to infringement.
Relevance:
Branding and trademark infringement in the NFT space may have significant licensing implications.
AI-generated NFTs that use or reference well-known brands could lead to similar challenges.
AI-created assets should avoid brand confusion and establish clear ownership and licensing terms to prevent infringement.
4. Conclusion
Licensing AI-generated art assets is a rapidly evolving area, influenced by traditional copyright and trademark principles, but modified to account for the unique nature of AI as a creator. Key strategies for licensing include:
Clearly defining ownership of the generated content (human vs. AI authorship).
Clarifying rights for modification and resale in NFT or virtual environments.
Establishing fair use or transformative use arguments when training on existing data.
Trademark protection and licensing in the digital world, especially for brand assets like virtual goods or avatars.
The legal landscape for licensing AI-generated assets continues to evolve, and case law will shape future strategies in this space.
RELATED Blog
comments