Who owns AI-generated content? Copyright laws explained (2025)

In 2025, the rapid expansion of artificial intelligence (AI) in creative industries has forced legal systems across the globe to confront a fundamental question: who owns AI-generated content? From AI-composed music to machine-written literature, the ambiguity around copyright ownership has become a pressing issue for businesses, creators, and policymakers alike. This article breaks down the current copyright laws in major jurisdictions, explores landmark cases, and analyses how businesses can protect their AI-generated work. Finally, it offers expert insights and predicts future legal shifts that could reshape intellectual property rights in the AI era.

The legal landscape: U.S., UK, and EU Copyright Laws

United States: the human authorship requirement

Under U.S. copyright law, the doctrine of human authorship remains a cornerstone principle. The U.S. Copyright Office (USCO) has consistently rejected copyright claims for works created solely by machines. The landmark Thaler v. Perlmutter case (2023) highlighted this stance. Stephen Thaler, the founder of Imagination Engines, attempted to copyright an image autonomously generated by his AI system, DABUS. The USCO denied the request, asserting that copyright can only be granted to works with human authorship, which was confirmed on March 18, 2025 by the appellate court as well.

However, hybrid works, i.e. those partially created by AI but involving substantial human input, are more ambiguous. In Zarya of the Dawn (2023), an AI-generated comic book was granted partial copyright protection because its human author demonstrated meaningful creative contribution in arranging and curating the text and arrangement.

United Kingdom: an evolving stance on AI authorship

In the UK, the amended Copyright, Designs and Patents Act 1988 (CDPA), provides some leeway for AI-generated content. Section 9(3) of the Act specifies that the author of a computer-generated work is the person who made the necessary arrangements, which is a provision unique to UK copyright law. It gives businesses and developers some claim over AI-generated outputs. However, the UK Intellectual Property Office (IPO) has indicated that this framework may be re-evaluated to reflect the increasing autonomy of generative models.

European Union: AI and collective rights

The EU’s legal framework has been influenced by the EU AI Act, which establishes guidelines for AI usage but does not yet directly address copyright ownership. However, existing copyright laws, including the Directive on Copyright in the Digital Single Market (2019), generally require human authorship for protection. The recent Getty Images v. Stability AI lawsuit, which is still ongoing, is expected to address pivotal issues concerning the scraping and use of copyrighted materials for AI training.

Similarly, in OpenAI v. Author’s Guild (2024), some authors accused OpenAI’s language model of using their copyrighted books for training without consent. The case highlighted the murky waters of data scraping and fair use, with experts predicting stricter regulations on AI training practices.

AI copyright risks for businesses

For businesses, the legal uncertainties surrounding AI-generated content present tangible risks:

• Unclear ownership: companies investing in AI-generated marketing or creative assets may face challenges in asserting ownership, making their content vulnerable to unauthorised use.
• Regulatory challenges: DeepMind v. ICO (2021, UK), while not a direct copyright case, highlighted issues around data usage and AI. The Information Commissioner’s Office (ICO) ruled that Google’s DeepMind had unlawfully used patient data to develop its AI-powered health app, Streams. Although it focused on data privacy, the case raised broader concerns about how AI companies handle proprietary or sensitive data (which could also influence future copyright disputes involving AI training datasets). This case shows how AI-related legal issues are evolving in the UK, even though copyright-specific cases are still emerging.
• Litigation Risks: as seen in the Stability AI lawsuits, companies that use AI models trained on copyrighted data without proper licensing face potential litigation. As AI becomes more sophisticated, the real debate will be over what is fair use and what is infringement. Courts will have to assess how much AI-generated content relies on protected material. On this point, you may have read about the warning letters sent by Sony Music to AI music start-up companies, which are part of a broader industry effort to protect copyrighted materials.
• Content devaluation: if AI-generated works cannot be copyrighted, they may be treated as public domain material, reducing their commercial value.

Protecting AI-generated work: best practices

To mitigate risks and protect AI-generated assets, businesses can adopt the following strategies:

• Hybrid human-AI collaboration: ensuring that human input plays a substantial role in content creation can strengthen copyright claims.
• Contractual safeguards: include explicit clauses in contracts that define ownership of AI-generated outputs.
• Licensing agreements: when using third-party AI models, secure licensing agreements that cover both input datasets and resulting content.

Predictions: future legal shifts

EU AI Act: expanding copyright protections

The EU’s AI Act, which came into effect in 2024, primarily regulates AI safety and transparency. However, experts predict that future amendments will introduce explicit copyright rules for generative AI. These may include requirements for AI models to disclose training data sources and acquire licenses for copyrighted content.

U.S. policy reform: clarifying AI authorship

In the U.S., the Copyright Office and Congress are under pressure to reform outdated copyright laws. Legal experts anticipate new policies clarifying the threshold of human involvement required for copyright protection.

International harmonization

Legal scholars predict growing calls for international harmonization of AI copyright rules. The World Intellectual Property Organization (WIPO) is already exploring global frameworks to standardize AI-related IP rights.

Conclusion

As AI continues to reshape creative industries, the legal frameworks governing copyright ownership are being put to the test. While current laws largely exclude AI-generated content from protection, upcoming legal reforms and landmark cases will likely redefine these boundaries. For businesses, proactive legal strategies and hybrid creative approaches are essential to safeguarding their AI-generated work. The next few years will determine whether AI can be a creative partner for the content industry without also becoming a legal minefield.