A March 18, 2025, US court ruling denied copyright for AI-generated art 1. This decision affirms that human authorship remains essential for creative works.
The US Court of Appeals for the D.C. Circuit delivered this significant decision 1. It focused on the copyrightability of AI-generated artwork 2. The court upheld an August 2023 district court ruling 2.
Dr. Stephen Thaler, a computer scientist, initiated the case 2. He sought copyright for "A Recent Entrance to Paradise" 2. This visual work was created by his "Creativity Machine" 2. Dr. Thaler explicitly named the AI as the sole author 1. He claimed ownership of the artwork himself 1.
The U.S. Copyright Office (USCO) denied his request in August 2019 2. They cited the absence of human authorship 2. Thaler's internal appeals also failed 2. He then challenged the refusal in federal court 3. Thaler argued the human authorship requirement was unconstitutional 3. He believed it lacked legal support 3.
A recent US court ruling denied copyright protection for AI-generated artwork, emphasizing human authorship as a core requirement for intellectual property. This decision profoundly impacts artists, developers, and the broader AI industry.
The Thaler v. Perlmutter case addressed the copyrightability of AI artwork 2. On March 18, 2025, the U.S. Court of Appeals for the D.C. Circuit issued its decision 1. This ruling upheld an earlier judgment from August 2023 by the U.S. District Court 2.
Dr. Stephen Thaler initiated the case, seeking copyright for his AI-generated visual work 2. The artwork was titled "A Recent Entrance to Paradise" 2. Thaler stated that his "Creativity Machine," an AI algorithm, was the sole author 2. He listed himself only as the owner in the copyright application 1.
The U.S. Copyright Office (USCO) denied Thaler's application in August 2019 2. They cited the absence of human authorship for their decision 2. Thaler then challenged this refusal in federal court 3. He argued the human authorship rule was unconstitutional 3.
The court upheld the USCO's denial 2. It stated that copyright protection requires human authorship as a "bedrock requirement" 3. The Copyright Act of 1976 demands that eligible works must originate from a human being 1.
The court's decision relied on several legal points. It focused on the inherent implications of the term "author" within the Copyright Act 4.
Statutory Interpretation
Historical Precedent
Incentive Theory
The court rejected Thaler's argument that the work-made-for-hire doctrine applied 3. This doctrine requires an initial human author for rights transfer 4. No copyright existed to transfer because the AI artwork lacked human authorship from its start 5.
Thaler also suggested he should be considered the author by making the AI 1. The court declined this argument 1. It was waived because Thaler did not raise it during the administrative process 1. Also, it contradicted his earlier claim that the AI was the sole author 1.
The recent ruling profoundly impacts both human artists and AI developers. It creates uncertainty for creators using AI tools . This decision also guides developers on future AI product strategies.
Artists using AI face new copyright questions. The court did not define the necessary human involvement for AI-assisted works . This leaves creators unsure about their intellectual property rights . Artists who heavily use AI may worry about their efforts . Their "prompt engineering" might not lead to protected output . Thaler argued that denying copyright would harm creativity . The court stated machines do not respond to such incentives . This creates a critical gap for human effort in AI creative processes.
Some developers continue pushing for AI authorship . Stephen Thaler's ongoing legal battle shows this intent . His counsel plans an appeal, calling the USCO policy "extra-statutory" . However, the court did not stop protection for works created by humans using AI . This suggests a path for AI tools that assist human creativity . The ruling highlights the need for legislative action by Congress . Expert agencies like the USCO should also address these issues . This might lead to more developer engagement in policy discussions.
Legal professionals widely expected this outcome . It aligns with existing U.S. copyright law . Kristelia Garcia from Georgetown University agrees with the court . She believes amendments are necessary to change the law . Edward Lee of Santa Clara University sees a larger battle ahead . This involves defining sufficient human contribution in AI-assisted works . He points to Allen v. Perlmutter as a key test for prompt engineering .
Shubha Ghosh from Syracuse University notes the complexity . It is hard to sort user prompt contribution from AI's creative output . AI's creativity can sometimes exceed the human author's . Ryan Abbott, Thaler's counsel, disputes the opinion . He argues the USCO policy violates the Copyright Act's purpose . Lutzker & Lutzker LLP advises clients on qualifying AI-assisted works 7. Skadden, Arps, Slate, Meagher & Flom LLP observes the ruling's limited scope 6. It does not fully address "thornier" questions of human creativity in AI-generated works 6. Businesses should consider trade secrets for purely AI-generated content 8.
Traditional intellectual property frameworks struggle to protect AI-generated content, forcing discussions around new legal and ethical protections for creators and developers 9. The US court's denial of copyright for AI artwork highlights these inadequacies, emphasizing the need for adaptable solutions 9. This rapidly changing landscape demands innovative approaches beyond existing copyright law 9.
Sui generis, meaning "of its own kind," represents a specialized legal framework. It offers protection distinct from traditional copyright 13. This framework is ideal for non-original digital content, particularly AI-generated works that lack human creative input 15. Its goal is to foster innovation and incentivize investment in AI technology 14.
Ukraine implemented a sui generis right on January 1, 2023 15. Article 33 of its law protects non-original objects generated solely by computer programs 15. The object must differ from existing similar items and not be a copy 13. Human involvement is limited to activating the software, not making creative decisions 17. Protection relies on expression form and substantial investment, rather than originality 15.
Economic rights are granted to the initiator or other legal rights holders of the AI system 13. These include reproduction and distribution rights 13. Moral rights are excluded, as outputs are not traditional "works" 13. Protection lasts for 25 years from the year following generation 17. This contrasts with traditional copyright's 70 years post-author death and its requirement for human originality and moral rights 17.
However, Ukraine's model faces challenges. Clarity is needed on autonomous generation, novelty, and rights holder identification 15. The EU maintains that authorship is exclusively human, limiting existing sui generis protection to database rights 15. Similarly, autonomously generated AI content in the US often lacks legal protection due to strict human authorship rules 15.
"Res Publicae ex Machina" proposes a new public domain model for highly autonomous AI creations 19. This term means "Public Property from the Machine" 19. This approach aims to revitalize the public domain. It makes purely AI-generated works freely available immediately, preventing private monopolies 19.
Proponents argue this is a "Pareto improvement" because many benefit without harming legal persons 19. They suggest extending copyrights to AI stifles innovation and cultural diversity 19. Such works could receive an official Public Domain (PD) mark 19. AI-assisted creations would typically not qualify for this automatic status 19. This model ensures immediate public use, diverging from traditional copyright's exclusive monopolies 19.
Hybrid or enhanced licensing models combine traditional copyright with AI-specific attribution and licensing 12. These models grant limited economic rights to AI developers or users. They avoid designating the AI as an author 14. Their purpose is to address blurred authorship in AI-human co-creation. They ensure fair ownership, promote legal certainty, and stimulate creative innovation 12. These models balance human-centric principles with protecting against economic loss from unprotected AI works 16.
Mechanisms for these models include bespoke AI agreements and co-creation frameworks 12. They would rely on transparent attribution, updated contract language, and shared royalty models 12. Technologies like blockchain and metadata tools could support this 12. These models might also include licensing platforms for AI training data 20. This helps address concerns over using copyrighted material without consent 12. Existing licensing agreements often lack explicit clauses for AI, creating ambiguity 12. WIPO and the EU are discussing these approaches 14.
Contract-based solutions offer private agreements to define ownership and usage rights 21. These provide clarity and allocate rights among parties involved in AI development and content generation 21. Employment contracts often assign IP rights to the employer 21. Licensing agreements specify ownership of AI-generated outputs, with users often seeking exclusive rights 21. Non-disclosure agreements (NDAs) with AI providers safeguard proprietary data 18. They ensure company inputs and outputs remain confidential 18. These are private arrangements, different from statutory public rights.
Trade secret protection guards valuable AI-generated data or processes through confidentiality 18. This offers a viable mechanism for AI outputs not eligible for copyright or patent protection 18. Trade secret law does not require human creation 18. It protects valuable information as long as its secrecy provides economic value 18. This framework uses strict internal confidentiality measures and need-to-know access controls 18. Its duration is indefinite, lasting as long as confidentiality is maintained 18. This differs from the fixed terms of patents and copyrights.
The UK Copyright, Designs and Patents Act 1988 (CDPA) offers a unique perspective. Section 9(3) states the author of "computer-generated works" is the person making the "arrangements necessary" 22. This model allows copyright protection for works created by computers. It attributes authorship to a human who facilitates the creative process 16. Ownership usually vests in the human owning or controlling the AI system, provided originality is met 21.
This model accommodates machine involvement. It focuses on the human's role in "arranging" the creation 22. This contrasts with strict human authorship rules in the US and EU 22. However, this provision predates modern AI systems. It struggles with highly autonomous AI, raising questions about originality 16.
The evolving AI landscape calls for enhanced transparency and international coordination 22. "Dynamic Ethical Frameworks" and "Global Fair AI Use" policies are crucial 23. The EU AI Act mandates transparency for AI training data 9. Proposals also suggest a registry for AI-generated objects to boost transparency 9.
Greater international coordination is vital to reduce legal uncertainty 22. It helps harmonize approaches to protecting AI-generated content globally 22. Navigating these complex protection landscapes requires robust tools and clear strategies. For creators and developers building AI applications, platforms like Atoms help bring ideas to life. Atoms (https://atoms.dev) is an AI app builder for solo founders. You can describe your idea and get a working app with auth, database, and payments. Over 500,000 users have explored its capabilities, including building an AI app from scratch [https://atoms.dev/usecases/ai-app-builder].
These discussions underline a critical balance. The goal is to incentivize AI innovation while protecting human creators and ensuring fair remuneration 24. The ease of AI content generation challenges traditional IP assumptions about scarcity and enforcement 25.