Should AI-Generated Works Receive Copyright Protection? Rethinking Human Authorship in the Era of Generative AI

1. Introduction

Generative artificial intelligence (AI) has complicated one of copyright law’s most basic assumptions: that a protected work can ultimately be traced to human creative activity. Generative systems can now produce literary text, images, music and other expressive material in response to user instructions, sometimes with little further human intervention. The resulting output may closely resemble a work created by a human author, yet resemblance alone does not answer the legal question of authorship.

This problem arises largely because core copyright instruments did not foresee works created by machines. The Berne Convention structures international copyright protection around authors and their literary and artistic works.[1] In the European Union, neither the Information Society Directive nor the Digital Single Market Directive establishes a specific regime determining authorship of AI-generated outputs.[2] Existing EU copyright doctrine instead relies heavily on originality, which the Court of Justice of the European Union (‘CJEU’) has defined through the concept of the author’s ‘own intellectual creation’.[3] In Painer, the Court further associated originality with the author’s ability to make free and creative choices and thereby stamp the work with a ‘personal touch’.[4]

The rise of generative AI raises an important question: should an output receive copyright protection where no human being exercised sufficient creative control over the expression ultimately produced?

Different jurisdictions have approached this question differently. In the United States, the Copyright Act has been interpreted as requiring copyrightable works to be authored, in the first instance, by a human being.[5] UK copyright law handles this differently by protecting ‘computer-generated’ works even without a human creator, treating the person who set up the process as the author.[6] Other jurisdictions have adopted different thresholds or alternative forms of protection, illustrating the growing divergence in approaches to AI-generated works.[7]

This article argues that output generated entirely by artificial intelligence is not eligible for copyright protection simply because it resembles human creations; protection is predicated on human intellectual effort. Works created with the assistance of artificial intelligence may be subject to copyright protection when a human has personally shaped the expression. The distinction here must hinge on whether a human carried out the project with the aid of artificial intelligence or a machine produced it entirely.

2. Human Authorship and Originality in Copyright Law

The case against ordinary copyright for fully autonomous AI output begins with originality and authorship. While the Berne Convention leaves the term ‘author’ without a strict definition, its protection consistently revolves around the rights of human creators.[8] Modern EU copyright legislation likewise regulates the rights attached to works without creating a separate category of AI authorship.[9] This leaves the CJEU’s originality jurisprudence particularly important.

In Infopaq International A/S v Danske Dagblades Forening case, the Court of Justice of the European Union decided that copyright only protects works that show the author’s “own intellectual creation.”[10] Later, in the Painer case, the Court explained that this standard is met when authors use their creative freedom to make “free and creative choices” that give the work a clear “personal touch.”[11] Although these cases do not mention generative AI, their emphasis on human involvement provides a clear way to judge if AI-generated works can be protected by copyright.

The European Parliamentary Research Service (EPRS) reached a similar conclusion in its December 2025 comparative briefing. It noted that EU law has no dedicated rules for AI-generated works, while existing CJEU rulings tie originality directly to an author’s intellectual effort, personal touch, and free, creative choices.[12] The briefing concluded that AI outputs with little to no human input are unlikely to qualify for copyright, though it found far less legal clarity when a person uses AI as a creative tool.[13]

This distinction is also visible in academic literature. Gaffar and Albarashdi identify authorship and originality as central problems in applying conventional copyright frameworks to AI-generated works.[14] Spindler similarly examines AI-generated products against the existing human-focused structure of European copyright and the competing interests of access, protection, and incentives.[15]  These analyses prove that the question cannot be answered simply by judging how novel, elaborate, or artistic the final piece appears.

In copyright law, it is important to distinguish between originality, novelty, and aesthetic value. Just because artificial intelligence creates something new or unique does not mean it is legally original. What matters is whether AI’s output comes from a meaningful creative decision-making process. This issue gets more complicated as AI moves from being just a tool for humans to shaping how a work is expressed.

3. AI-Assisted Creation and Autonomous AI Generation

Copyright rules should distinguish between works made with help from artificial intelligence and those made entirely by machines. The main difference is about function, not just technology. Using AI in a project alone should not prevent protection. What matters most is whether a human truly shaped the form of expression that needs protection.

The United States Copyright Office (USCO) has created a detailed framework for copyright and AI. In its January 2025 report, Copyright and Artificial Intelligence, Part 2: Copyrightability, the Office found that using AI as a tool does not stop a work from being protected by copyright. If a person’s own expression is clear in the output, or if they make creative changes to AI-generated material, or show enough creativity in how they select, organize, or arrange content, the work may qualify for protection.[16] The Office also pointed out that, with current technology, prompts by themselves do not give users enough control to be considered the authors of the final output.[17]

The USCO’s Zarya of the Dawn decision highlights this distinction. Kristina Kashtanova’s graphic novel combined her own writing with images made using Midjourney. The Office decided that Kashtanova was the author of the text and the way the work was put together, but not of the individual Midjourney images.[18] This means that AI-generated parts might not be protected, while the human-created parts can still get copyright.

This principle can serve as the basis for a broader test of human creative control. Fritz builds a four-step test for AI-assisted works out of European case law. The analysis looks at the persons involved, the nature of the AI tool, and whether the user maintained real creative judgment and control over the final output.[19] Bektaş’s recent ‘Algorithmic Authorship Threshold’ takes a similar path, focusing on what counts as a legally meaningful human contribution when creators use AI under EU copyright rules.[20]

These approaches work better than treating the sheer number or complexity of prompts as the deciding factor. Entering a hundred prompts means little if the user still cannot control how the system turns those ideas into expressive form. Creators who actively filter, edit, combine, and arrange AI-generated material are making real creative choices that warrant protection.

The main requirement should be about quality. Is the final protected work the result of a human creator’s free and creative choices?

4. Comparative Approaches to AI-Generated Works

4.1 European Union

EU law currently contains no specific copyright regime for AI-generated outputs. The AI Act does not resolve AI authorship or define how much machine involvement is compatible with a human creator. Courts and practitioners must therefore look to the existing copyright acquis, which still operates on established principles of originality.[21] On 10 March 2026, the European Parliament adopted a resolution confirming the necessity of human authorship, pointing back to the CJEU rule that any work must represent the author’s own intellectual creation.[22]

Recent changes in national law are heading in the same direction. Italy announced that starting in 2025, works made with artificial intelligence may be protected by copyright if they show human intellectual effort. In the Czech Republic, a court decided that simply entering a command into an image generator usually does not count as authorship.[23] Although the EU has not set a common standard yet, these examples highlight how important clear human creativity is.

4.2 United States

The United States has clarified the “human authorship” requirement for copyright. In the case of Thaler v. Perlmutter, the D.C. Court of Appeals affirmed that copyright protection could not be granted to the work titled A Recent Entrance to Paradise, which was created without human intervention by Thaler’s system known as the “Creativity Machine.” The court stated that only works created by humans are eligible for copyright protection.[24]

The decision does not exclude AI-assisted work. In line with the US Copyright Office’s 2025 approach, copyright can protect identifiable human expression, creative modifications, or the careful selection and arrangement of AI-generated content. Prompting alone remains insufficient under today’s standard technology.[25] The US approach accommodates AI participation while keeping human authorship at the core of copyright protection.

4.3 United Kingdom

The United Kingdom provides a statutory contrast. Sections 178 and 9(3) of the CDPA recognize works that have no human author as ‘computer-generated’. Authorship is then attributed to whoever makes the arrangements necessary to create the work.[26] Bond and Blair describe the resulting conceptual tension as ‘authorship without an author’.[27]

This exceptional regime is now under reconsideration. The UK Government’s March 2026 Report on Copyright and Artificial Intelligence questions whether the original investment rationale still holds weight now that computer-generated works can be produced cheaply and on a scale. The report also notes that 78 percent of consultation respondents opposed keeping this protection, agreeing that entirely AI-generated works should not qualify for copyright.[28] The UK’s track record cautions against automatically granting standard copyright protection to autonomous AI works.

4.4 China and Ukraine: Alternative Models

China takes a case-by-case approach that focuses on specific facts. Chinese courts have granted protection when there is clear evidence of personalized human creativity, such as detailed prompt inputs, parameter changes, and editing. In other cases, courts have required stronger proof that a person intentionally controlled the creative elements.[29]

Ukraine took a different approach by creating special rights for certain AI-generated works, while keeping regular copyright only for human authors.[30] This means machines cannot be considered authors, but it remains unclear whether AI-generated works deserve their own exclusive rights.

5. Should Fully AI-Generated Works Receive Copyright Protection?

The main reason to support protection is economics. Generative AI systems require significant investment, and granting exclusive rights to their outputs could encourage more development and commercialization. Bonadio, McDonagh, and Dinev describe this debate as a tension between models focused on creativity and those based on investment.[31] Feldman also asks whether traditional intellectual property rules, which were designed for human authors and original work, can include AI-generated creations.[32]

The UK’s historical regime reflects this investment rationale. As the 2026 Government report explains, the 1988 provisions on computer-generated works were meant to encourage capital investment in developing AI technologies.[33] That justification carries far less weight today, as the report itself acknowledges that computer-generated works can now be produced cheaply and in vast quantities.[34] At a fundamental level, financial investment does not equate to authorship or original expression. Capital expenditure alone fails to justify standard copyright protection.

Sometimes, automated protection systems can be overly strict. Generative systems can quickly produce large amounts of impressive content at a low cost. If traditional copyright rules were applied to all machine-generated output, this could expand the scope of exclusive rights without fostering genuine human creativity. An EPRS briefing note points out a similar risk, warning that protecting content not made by humans could increase the power of automated platforms and make human-created works less visible and valuable.[35]

In 2025, the USCO came to a similar policy decision. After reviewing other copyright and sui generis options, the report found insufficient evidence to support new exclusive rights. Developers already have strong incentives, including patent and trade secret protection for their technologies, as well as good funding and the benefits of being first to market. The rapid growth of AI adoption also weakens the argument that outputs need extra legal protection to encourage innovation.[36]

If future evidence shows a real market failure, a unique legal approach could avoid changing traditional copyright rules. Ukraine already offers a working example. Still, just because this is legally possible does not mean it is needed. New monopolies should only be granted to solve proven economic problems, not just because machines can create valuable content.

A more effective way is to use a human creative control standard. AI-assisted works should only get copyright protection if the original expression comes from the creator’s own free and creative choices. This can be shown through human-written sections, creative edits, choices in how things are selected and arranged, or real control over the finished product. Just giving a prompt and accepting whatever the machine makes is not enough. This approach keeps the focus on whose creative decisions shape the work, not just the tool they use.

6. Conclusion

Generative AI has sharply highlighted the fine line between outputs that “appear creative” and works that -in the eyes of the law- embody genuine human effort and creativity. Although different countries approach the issue from varying perspectives, copyright laws in both the European Union and the United States continue to regard the human factor as the fundamental criterion. Even the United Kingdom, which previously afforded special protection to computer-generated works, is now revisiting this approach.

The fact that AI outputs can be complex or valuable is not enough reason to give them standard copyright protection. If we grant protection in these cases, we move away from rewarding human creativity and create more exclusive rights without showing they are needed to encourage progress.

If a creator still has creative control over their work, using AI tools should not stop them from getting copyright protection. The main question should be about the human’s creative input, not the tool they used. Copyright is meant to reward human creativity that uses technology, not machine-made works that only imitate human art.

Reference(S):

Cases and Decisions

Infopaq International A/S v Danske Dagblades Forening (Case C-5/08) EU:C:2009:465.

Painer v Standard VerlagsGmbH and Others (Case C-145/10) EU:C:2011:798.

Thaler v Perlmutter 130 F 4th 1011 (DC Cir 2025).

US Copyright Office, Zarya of the Dawn (Registration No VAu001480196, 21 February 2023).

Legislation and International Instruments

Berne Convention for the Protection of Literary and Artistic Works (Paris Act 1971, amended 28 September 1979).

Copyright, Designs and Patents Act 1988.

Directive 2001/29/EC of the European Parliament and of the Council of 22 May 2001 on the harmonisation of certain aspects of copyright and related rights in the information society [2001] OJ L167/10.

Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital Single Market [2019] OJ L130/92.

EU Materials and Official Reports

Department for Science, Innovation and Technology, Department for Culture, Media and Sport and Intellectual Property Office, Report on Copyright and Artificial Intelligence (March 2026).

European Parliament resolution of 10 March 2026 on copyright and generative artificial intelligence – opportunities and challenges (2025/2058(INI)) [2026] OJ C 4006.

Karttunen S, Copyright of AI-Generated Works: Approaches in the EU and Beyond (European Parliamentary Research Service, December 2025) PE 782.585.

US Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (Report of the Register of Copyrights, January 2025).

Books and Book Chapters

Bonadio E, McDonagh L and Dinev P, ‘Copyright in Works Created by Artificial Intelligence: Between Creativity and Investments’ in Enrico Bonadio and Patrick Goold (eds), The Cambridge Handbook of Investment-Driven Intellectual Property (Cambridge University Press 2023).

Feldman R, ‘IP for AI Creations?’ in AI versus IP: Rewriting Creativity (Cambridge University Press 2025).

Spindler G, ‘AI and Copyright Law: The European Perspective’ in Larry A DiMatteo, Cristina Poncibò and Michel Cannarsa (eds), The Cambridge Handbook of Artificial Intelligence: Global Perspectives on Law and Ethics (Cambridge University Press 2022).

Journal Articles

Bektaş UG, ‘The Algorithmic Authorship Threshold: Identifying Human Contribution in AI-Assisted Creativity’ (2026) 21 Journal of Intellectual Property Law & Practice 610.

Bond T and Blair S, ‘Artificial Intelligence & Copyright: Section 9(3) or Authorship without an Author’ (2019) 14 Journal of Intellectual Property Law & Practice 423.

Fritz J, ‘Understanding Authorship in Artificial Intelligence-Assisted Works’ (2025) 20 Journal of Intellectual Property Law & Practice 354.

Gaffar H and Albarashdi S, ‘Copyright Protection for AI-Generated Works: Exploring Originality and Ownership in a Digital Landscape’ (2025) 15 Asian Journal of International Law 23.

[1] Berne Convention for the Protection of Literary and Artistic Works (Paris Act 1971, amended 28 September 1979) arts 1–2.

[2] Directive 2001/29/EC of the European Parliament and of the Council of 22 May 2001 on the harmonisation of certain aspects of copyright and related rights in the information society [2001] OJ L167/10; Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital Single Market [2019] OJ L130/92.

[3] Case C-5/08 Infopaq International A/S v Danske Dagblades Forening EU:C:2009:465, paras 37–39.

[4] Case C-145/10 Eva-Maria Painer v Standard VerlagsGmbH and Others EU:C:2011:798, paras 87–92.

[5] Thaler v Perlmutter 130 F 4th 1011 (DC Cir 2025).

[6] Copyright, Designs and Patents Act 1988, ss 9(3), 178.

[7] Sofia Karttunen, Copyright of AI-Generated Works: Approaches in the EU and Beyond (European Parliamentary Research Service, December 2025) PE 782.585, 3–5.

[8] Berne Convention (n 1) arts 1, 2 and 5.

[9] Directive 2001/29/EC (n 2); Directive (EU) 2019/790 (n 2).

[10] Infopaq (n 3) paras 37–39.

[11] Painer (n 4) paras 87–92.

[12] Sofia Karttunen, Copyright of AI-Generated Works: Approaches in the EU and Beyond (European Parliamentary Research Service, December 2025) PE 782.585, 2.

[13] Karttunen (n 12) 2.

[14] Hafiz Gaffar and Saleh Albarashdi, ‘Copyright Protection for AI-Generated Works: Exploring Originality and Ownership in a Digital Landscape’ (2025) 15 Asian Journal of International Law 23, 23–46.

[15] Gerald Spindler, ‘AI and Copyright Law: The European Perspective’ in Larry A DiMatteo, Cristina Poncibò and Michel Cannarsa (eds), The Cambridge Handbook of Artificial Intelligence: Global Perspectives on Law and Ethics (Cambridge University Press 2022) 257–270.

[16] US Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (Report of the Register of Copyrights, January 2025) 27, 41.

[17] ibid 18–20.

[18] US Copyright Office, Zarya of the Dawn (Registration No VAu001480196, 21 February 2023) 1, 9–12.

[19] Johannes Fritz, ‘Understanding Authorship in Artificial Intelligence-Assisted Works’ (2025) 20 Journal of Intellectual Property Law & Practice 354, 354–364.

[20] Ulaş Göker Bektaş, ‘The Algorithmic Authorship Threshold: Identifying Human Contribution in AI-Assisted Creativity’ (2026) 21 Journal of Intellectual Property Law & Practice 610, 610–618.

[21] Sofia Karttunen, Copyright of AI-Generated Works: Approaches in the EU and Beyond (European Parliamentary Research Service, December 2025) PE 782.585, 2; Directive 2001/29/EC (n 2); Directive (EU) 2019/790 (n 2).

[22] European Parliament resolution of 10 March 2026 on copyright and generative artificial intelligence – opportunities and challenges (2025/2058(INI)) [2026] OJ C 4006.

[23] Karttunen (n 10) 2–3.

[24] Thaler v Perlmutter 130 F 4th 1011 (DC Cir 2025).

[25] US Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (Report of the Register of Copyrights, January 2025) 18–20, 27, 41; Thaler (n 24).

[26] Copyright, Designs and Patents Act 1988, ss 9(3), 178.

[27] Toby Bond and Sarah Blair, ‘Artificial Intelligence & Copyright: Section 9(3) or Authorship without an Author’ (2019) 14 Journal of Intellectual Property Law & Practice 423.

[28] Department for Science, Innovation and Technology, Department for Culture, Media and Sport and Intellectual Property Office, Report on Copyright and Artificial Intelligence (March 2026) Section I, paras 3–6, 31–34.

[29] Karttunen (n 10) 3–4.

[30] Karttunen (n 10) 4–5.

[31] Enrico Bonadio, Luke McDonagh and Plamen Dinev, ‘Copyright in Works Created by Artificial Intelligence: Between Creativity and Investments’ in Enrico Bonadio and Patrick Goold (eds), The Cambridge Handbook of Investment-Driven Intellectual Property (Cambridge University Press 2023) 73–85.

[32] Robin Feldman, ‘IP for AI Creations?’ in AI versus IP: Rewriting Creativity (Cambridge University Press 2025) 61–67.

[33] Department for Science, Innovation and Technology, Department for Culture, Media and Sport and Intellectual Property Office, Report on Copyright and Artificial Intelligence (March 2026) Section I, para 3.

[34] Department for Science, Innovation and Technology and others (n 33) Section I, paras 5–6.

[35] Sofia Karttunen, Copyright of AI-Generated Works: Approaches in the EU and Beyond (European Parliamentary Research Service, December 2025) PE 782.585, 4–5.

[36] US Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (Report of the Register of Copyrights, January 2025) 36–37.