Skip to Main Content

Generative Artificial Intelligence (GenAI): Copyright

Generative Artificial Intelligence (AI) Guide

AI and Copyright Issues

Artificial Intelligence (AI) has exploded in the social mainstream with the launch of ChatGPT by OpenAI. This has led to some countries embracing the new technological dawn, while others are banning it. There are multiple legal and ethical issues involving the use of AI, as much as there are benefits. Copyright law and AI is only one branch of this new evolving field, and understanding how copyright plays into the use of AIs is important for a compliant use of these technologies in a Higher Education environment.

Copyright in Generative AI:

  • Training Data and Copyright:

    • Generative AI models are often trained on large datasets that may include copyrighted material.
    • Where a generative AI model outputs its own copyrighted training data, this will likely infringe copyright. More broadly, outputs which resemble the training data may infringe copyright depending on the copyright status of the training data, the terms under which its use was licenced and specific copyright legislation in the country where the response was output. Note that in many countries the law regarding AI is unclear and may be affected by pending legislation.

  • Output Ownership:

    • In almost all jurisdictions, AI-generated content without human creative input may not qualify for copyright protection.
    • If a human significantly guides or edits the output, they may hold copyright over the final work.

  • Infringement Risks:

    • Some prompts may cause AI systems to create outputs which infringe copyright. For example, if asked to depict an Italian plumber, many image generation systems used to generate images of Nintendo’s Mario until guard-rails were created in the system to prevent this. Deliberately trying to get a system to create such images is copyright infringement.
    • Many AI systems ingest the data used in their prompts into their own training data, unless the user specifically selects an option to ensure that they do not do this. If copyrighted material is used as part of a prompt and the system reproduces it into the training data, this may constitute copyright infringement.
    • This raises concerns about plagiarism or unauthorized reproduction.

  • Licensing and Usage Rights:

    • Some AI tools specify in their terms of service who owns the generated content and what rights users have.
    • Always check the license agreement of the AI platform you're using.

  • Moral and Ethical Considerations:

    • Even if legally permissible, using AI to mimic an artist’s style can raise ethical concerns.

  • Legal Uncertainty:

    • Copyright law is still evolving to address the challenges posed by generative AI.
    • Different countries may interpret and enforce these laws differently.
    • If you require any further assistance, guidance or support, please contact: copyright@ulster.ac.uk

JISC Guidance

 

 

 

 

 

 

Jisc has issued guidance on Generative AI and copyright law and practice in education (March 2024) which raises some 'challenging questions' as a starting point for navigating copyright issues:

▷ If I use generative AI tools to generate content, am I at risk of infringing the rights of someone whose work was used to train the model?

▷ Will I as a user of the tool own the work that is created by my prompts? Or do authors whose works are used to train AI systems have an ownership claim? Or is the AI tool the author?

▷ Does the material that I input to a generative AI tool become part of the Generative AI tool database for others to use?

▷ If I input someone else’s content into a generative AI tool and content is generated could this be an infringement of their copyright?

Additional Information

Copyright Ownership

Understanding copyright in the context of Generative AI (GenAI) involves knowing who owns the rights to prompts, outputs, and reused research. Ownership and protection depend on originality, jurisdiction, and the terms of the AI tool used. Legal interpretations vary globally, and users must be cautious about how their content is used and reused.

🔹 Prompts You Provide

  • Prompts you create may be protected by copyright if they are original.
  • AI tools may reuse your prompts for training unless you opt out.
  • Always check the tool’s terms and conditions for data usage and licensing.
  • Avoid inputting confidential or commercially sensitive information.

🔹 Reuse of Research Publications

  • Open access research may be used to train GenAI models under existing licenses.
  • Subscription content may be used if lawful access and copyright exceptions apply.
  • Concerns include attribution, derivative works, and copyright infringement.
  • Publishers may seek permission to use your work in training datasets—check what’s covered and ask about:
    • Purpose of reuse
    • Future control
    • Attribution
    • Royalties (especially for books)

Copyright Guidance

This guidance gives a brief overview of UK copyright law and how it affects you, explains how to carry out common tasks to support teaching without infringing copyright and contains a set of FAQs answering common queries about copyright.

The information given here should not be construed as legal advice. Should you require legal advice, please contact either the Legal Services department or another qualified legal professional.

This guide is not exhaustive and does not address many aspects of copyright law. If you have a question about copyright which is not addressed here, please contact the University’s Copyright Officer.

Copyright Law and Gen AI: a legal framework

Copyright definition states that for a work to be protected by copyright, it needs to be created by a ‘natural person’. However, UK law S.9(3) of the UK Copyright Designs and Patents Act 1988 (CDPA) states that ‘(3) In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.’

The wording here does not fully define ownership. ‘The person by whom the arrangements necessary for the creation of the work are undertaken’ is not clear in a scenario in which a user provides prompts for the AI generator to produce outputs. There has also been little by way of case law to assess the meaning of the legislation as it applies to AI.

Some countries around the world have taken measures to create a legal framework in the use of AIs: 

  • European Union: AI Act focuses on risk mitigation. It includes transparency requirements for AI used in training models, potentially impacting copyright issues by demanding of disclosure of copyright materials used;
  • China: has several guidelines and regulation which aim to balance innovation with copyright protection;
  • South Korea: AI Basic Act has been enacted in 2020, promoting responsible AI use but not specifically targeting copyright issues;
  • Singapore: issued guidelines on ethical AI development, including data collection and sharing, which could have implications for copyright.

The rapid evolution of AI has outpaced legal frameworks. Legislation addressing the ethical and copyright implications of AI varies widely between countries. While some jurisdictions focus primarily on ethical considerations, others encompass both ethical and copyright issues.