The Question Reshaping Creative Industries
Generative AI tools can now write novels, compose music, generate photorealistic images, and produce functional software code. As these tools become embedded in professional workflows, a foundational legal question has moved from theoretical to urgent: Who, if anyone, owns the copyright in content generated by AI?
The answer — at least for now — is deeply unsettled, and the legal landscape is evolving rapidly across the United States, European Union, and other major jurisdictions.
The Human Authorship Requirement
In the United States, the Copyright Act and decades of case law have consistently held that copyright can only protect works created by human authors. This principle was reinforced in a series of rulings involving AI-generated art submitted by researcher Stephen Thaler on behalf of an AI system called DABUS.
In those cases, US courts and the Copyright Office consistently ruled that works produced autonomously by AI — without sufficient human creative input — are not eligible for copyright protection. The Copyright Office has issued guidance stating that it will not register works "produced by machine or mere mechanical process" that lack human authorship.
This creates an important practical gap: an AI-generated image, article, or piece of music may be freely copied by anyone, since it belongs to no one — it falls into the public domain from the moment of creation.
The "Human Authorship" Spectrum
The more nuanced question is: how much human input is enough to qualify for copyright protection? The Copyright Office has indicated it will evaluate AI-assisted works on a case-by-case basis, looking at the degree of human creative control. Consider three scenarios:
- Fully autonomous AI output: A user types a one-word prompt and accepts the AI's output unchanged. Courts and the Copyright Office have found this insufficient for copyright protection.
- Heavily curated and edited AI output: A human artist uses AI as a tool, then substantially modifies, selects, arranges, and layers content. The human-contributed elements may be protectable.
- AI as a tool (like Photoshop): A human directs an AI to execute a specific, detailed creative vision with precise control over the output. This is more analogous to using software as a tool and more likely to support a copyright claim.
The line between these categories is blurry, and litigation will continue to refine it over coming years.
What About the Training Data?
A separate but related controversy involves whether training AI models on copyrighted works constitutes infringement. Several major lawsuits are currently working through US and international courts, with plaintiffs including authors, visual artists, and news organizations arguing that AI developers violated their copyrights by ingesting their works without permission or compensation.
Key legal questions in these cases include:
- Does training an AI on copyrighted data constitute a "copy" under copyright law?
- Does the fair use doctrine shield AI developers from liability for training data?
- Are AI-generated outputs "derivative works" of the training data?
These questions are being litigated globally, and outcomes will vary by jurisdiction. The EU's AI Act and existing copyright directive create a different framework than US law, particularly around text and data mining exceptions.
The EU Approach
The European Union's AI Act, which entered into force in 2024, requires providers of general-purpose AI models to publish sufficiently detailed summaries of the content used in training. The EU Copyright Directive also contains opt-out mechanisms allowing rights holders to prevent their works from being used in AI training. While the EU has not explicitly addressed AI authorship in the same way as the US, the general consensus in European copyright law also requires human creativity for protection.
Practical Implications for Businesses Using AI Tools
- Don't assume AI output is protected: If you publish AI-generated content without meaningful human creative input, competitors may freely reproduce it.
- Document your creative process: Keep records of human decisions, edits, and creative choices made during AI-assisted production to support future copyright claims.
- Review AI vendor terms: Understand what rights, if any, the AI platform claims over outputs and what indemnification they offer against third-party infringement claims.
- Monitor litigation developments: Court decisions in pending AI copyright cases will significantly shape what's legally permissible.
Conclusion
Copyright law, built on a century of human creativity assumptions, is straining under the pressure of generative AI. Businesses and creators using AI tools need to understand that the legal protections they expect may not apply — and that the rules are actively being rewritten by courts, regulators, and legislators worldwide.
This article is for informational purposes only and does not constitute legal advice. Consult a qualified attorney for guidance specific to your situation.