Mum's the word – generative AI and copyright infringement

It has been reported that Mumsnet, a parents networking website, is taking legal action against OpenAI. Mumsnet objects to OpenAI "scraping" the Mumsnet website, in order to train its large language model (LLM) and ultimately facilitate its AI tool, ChatGPT.

The Mumsnet founder has stated that this is an explicit breach of copyright and the website's terms of use. We have blogged previously about other copyright actions taken against OpenAI, but this case is the first of its kind in the UK.

This litigation is just one among a growing number of cases worldwide where companies and content creators are beginning to challenge how AI systems are developed, spotlighting the evolving legal landscape around generative AI and intellectual property (IP) rights.

What is generative AI?

Generative AI is a machine-learning model that is trained to create new data based on the data which is input into the system. The more data it receives, the more it learns from itself. As our Innovation and Technology Director has written on previously, generative AI and the LLMs which they are based on have the potential to change work, sectors and businesses in fundamental ways.

ChatGPT is a form of generative AI. It is a chatbot and virtual assistant which allows users to have conversations in order to complete various tasks, such as drafting articles and scripting social media posts.

Generative AI, while powerful, relies heavily on large-scale datasets—often scraped from publicly accessible websites—which raises complex questions about ownership and the legality of using such data without permission. From a legal perspective, this poses significant challenges for existing IP frameworks, such as copyright law, which were not designed to handle these new technologies.

What are the IP risks?

As generative AI does not credit the original creators of the data it gathers, there is a growing concern from the creators of artistic, musical and literary content that this poses a fundamental risk of widespread copyright infringement.

In other jurisdictions, it is becoming increasingly common to see creative rights holders raising legal proceedings for copyright infringement in the context of generative AI. For example, a group of eight US newspapers including The New York Daily News and Denver Post, have raised an action against Microsoft and OpenAI alleging that they have "illegally harvested millions of copyrighted articles" to create products including ChatGPT and Microsoft’s Copilot.

However, the Mumsnet litigation is the first we have seen in the UK. Mumsnet's claims include copyright infringement, breach of terms of use, and database right infringement. It is also seeking an order for OpenAI to be required to delete all Mumsnet data it currently holds and to cease any future use of the data. Mumsnet has reported that it approached OpenAI to discuss the possibility of entering into a licensing agreement, on the basis that "AI models have misogyny baked in and [Mumsnet would] love to help counter the gender bias likely to be present in many of them and raise women’s voices", but no agreement was reached.

For copyright owners, the difficulty with seeking to agree licences and/or taking court action is that often, by the time they become aware of the alleged infringement, the material has already been used on several different AI platforms and control of it is somewhat lost.

This poses a fundamental question – how can owners of creative works preserve the integrity of their work and retain control of their IP? The key legal issue here revolves around whether training an AI on copyrighted data constitutes a breach of copyright law, a topic that UK courts may need to interpret in light of the Mumsnet case. The Copyright, Designs and Patents Act 1988, which governs much of UK copyright law, was not designed with AI in mind, potentially leaving a legal grey area that courts will need to address.

Future proofing against the risks of AI

When delivering its response to the AI Regulation Consultation, the UK Government confirmed in February 2024 that it was committed to supporting both the AI technology sector and creative sectors but that it had become clear that, at least at present, a voluntary code of practice could not be agreed amongst those in the respective industries.

Nevertheless, the Government did state that it would endeavour to ensure that there is a future mechanism put in place to allow creators of content to establish what AI developers have used to train their AI products. There is a view that in order to forge a productive way forward there will require to be significantly more transparency from AI developers as to the material and technology used to underpin their AI technology.

The outcome of the Mumsnet case may well be a useful indication as to how the law is likely to develop in this area going forward.

One potential solution for rights holders may be to embed technological safeguards within their content, such as watermarking or digital rights management (DRM) tools, to prevent unauthorised use by AI systems. At the legislative level, the EU’s AI Act could serve as a model for how to regulate AI usage in ways that protect IP rights while fostering innovation.

Conclusion

Generative AI has extraordinary capabilities and the exponential growth in its user numbers is testament to that fact. However, as the dispute between Mumsnet and OpenAI shows, the development of AI may pose risks to the ability of businesses and individuals alike to protect their important IP rights.

As we have blogged previously, when considering deploying or utilising ChatGPT or other AI systems in your organisation, IP issues are just one risk to consider. As part of any risk assessment, organisations should consider the risks associated with each use case. We are continuing to work with clients to help them develop AI playbooks and guidance.

Given the legal uncertainty surrounding AI and copyright, businesses should consider reviewing their data policies and establishing clear guidelines for the use of AI-generated content. Proactively securing licensing agreements or implementing stricter terms of use may provide a buffer against future legal challenges, while also contributing to the broader discourse on AI ethics and governance.

If you would like to speak to one of our experts about protecting your business in the age of AI, or IP strategies in general, please get in touch with a member of the IP, Technology & Data team or your usual contact at Brodies.