Who Owns AI-Created Content? What Businesses Need to Know About Copyright and Generative AI

Why It Matters

As generative AI becomes a coveted tool in business, the legal ownership of its outputs remains unclear. Understanding where copyright applies – and where it doesn’t – is critical to protect creative assets and avoid legal risks.

Key Takeaways

• AI-generated content isn’t protected by copyright in most countries, unless substantial human input is involved, and even so, there’s no guarantee that it attracts copyright protection.
• Copyright laws vary across jurisdictions, and businesses must tread carefully depending on where and how IP content is used.
• Firms can still harness value from AI outputs, but they must know how to protect and commercialise them within legal boundaries.

Do You Have Rights Over What AI Systems Creates? It Depends

Generative AI tools, from ChatGPT to Midjourney, are transforming how businesses create content, from designs, music, to company logos. But who owns the rights to this AI-generated material? In most jurisdictions, including Australia, Singapore, and the United States, the answer hinges on several factors, the most important being: human input.

Copyright law protects the expression of ideas – not the ideas themselves – and it requires a human author. Content created entirely by AI systems, without meaningful human involvement, usually falls outside legal protection. Courts and Copyright Offices in the US and Singapore have rejected claims over AI-generated works when there’s no demonstrable intellectual contribution from a person.

Even in the UK, where the law is more flexible and acknowledges computer-generated works, copyright law still requires that someone, usually the person making the creative arrangements, be identifiable as the author. Still, the law predates modern AI and is grounded upon centuries of sound legal jurisprudence. This legal grey area leaves businesses exposed unless they actively shape, adapt, or refine the output of AI systems in a calibrated way.

What Counts as Original? Varying Legal Requirements Around the World

Copyright protection only subsists if a work is “original”, but how that is judged varies by country, and the legal situation differs in Singapore, Australia, UK and EU as analysed in the article.

The US Copyright Office has issued Guidance that AI-assisted works may satisfy the originality threshold and qualify for protection, but only if the user can show sufficient human authorship as required under the law. Even then, only the human-authored parts are protected, not the AI- generated material itself.

Legal Risks and Workarounds for Businesses

Beyond copyright ownership issues, businesses also face the risk of infringement. Generative AI systems are often trained on massive datasets scraped from the internet, including copyrighted works. There have already been lawsuits from publishers and image libraries (like Getty Images and The New York Times) claiming their content was used without permission by generative AI system developers.

There are key strategies that organisations can adopt which are explored in the chapter.

Business Implications

Be Strategic with AI Use: Don’t assume that content created by AI belongs to your company by default. Ensure team members understand when human input is needed to secure copyright protection, especially in marketing, branding, and product development.

Avoid Infringement: AI tools can unintentionally replicate copyrighted material. Use them with caution, especially for external-facing assets, and invest in content audits or legal reviews where appropriate.

Build Copyright-Eligible Assets: Where possible, train staff to add value to AI outputs. These contributions help convert AI-generated content into proprietary assets that your organisation can confidently use, protect and monetise.

 

Author and Sources

Author: Hannah Yee-Fen Lim (Nanyang Technological University)

Original publication: Data Strategy and AI Value Creation (2025), Chapter 13