To Regulate or to Innovate? Exploring Chatbots, Copyright and Creation
31/03/25We are excited to introduce a new Cambridge Mind Technologies blog series, entitled Unleashing AI: Legal Perspectives on Regulation.
The legal impact of artificial intelligence will span across all industries and sectors, leaving no stone unturned. The current rate of digital development far outpaces the speed at which carefully considered legislative amendments can be discussed and implemented. Governments across the globe must prioritise effectively and target key issues to adapt current frameworks. In navigating these unfamiliar legal waters, the UK government launched the AI Opportunities Action Plan in January 2025, with the hope of positioning the UK as an AI superpower. However, ‘turbocharging AI’¹³ will require us to address the ways in which our existing laws are inadequate. This blog series will explore the proposed strategies to accelerate AI innovation and examine their impact on legal precedent.
The first blog of this new series - To Regulate or to Innovate? Exploring Chatbots, Copyright and Creation - examines the UK government proposals for a novel exception in copyright infringement, and the careful balance that must be found in order to support both AI developers and IP rights-holders:
The Key Tension
The recent boom in AI technology, particularly the advancements in chatbots and large language models’ (LLMs) capabilities, has no doubt been made possible due to the vast swathes of internet data, in quantities unimaginable to developers 25 years ago. From social media to digital libraries and online resources, these new tools can tap into an almost endless bank of images and text at the click of a button. Widespread text and data mining (TDM) practices have raised novel issues surrounding whether such actions amount to intellectual property (IP) infringement. The lack of clear regulations surrounding the use of TDM for training AI models has created undesirable confusion in the field. As data mining becomes increasingly important for the creation of AI tools, policymakers must prioritise finding a solution which supports AI innovation while ensuring adequate protection for IP rights holders.
The IP implications of AI tools stretch far beyond LLMs and AI developers themselves, holding keen importance for digital creators, startups, and businesses looking to leverage this emerging technology. For Cambridge Mind Technologies, and other companies looking to harness the potential of AI to create robust and innovative tools, the uncharted legal territory surrounding TDM is a fascinating and crucially important space to engage with.
Proposed Responses
So far, the tech and AI industry’s mantra in this regard has been ‘better to ask for forgiveness than permission’. However, as much as they may benefit from a mostly unregulated space, any uncertainty, in turn, opens developers up to potential liability. As TDM becomes widespread, many creators and rights holders have taken the protection of their IP into their own hands. For instance, they are utilising platforms such as robots.txt or ‘Have I Been Trained?’ to signpost to developers that certain sites are off-limits¹. The UK government, prior to launching the AI Opportunities Action Plan² in January 2025, had grown increasingly concerned that the legal uncertainty in this field was discouraging innovation, and in hopes of attracting more home-grown AI projects, launched a consultation (in December 2024), calling for public feedback on their proposals for reforming copyright law in light of AI developments³.
Current defences against copyright infringement are wider than those of other IP rights, but do adequately address the situation of TDM. Previously, the government proposed introducing a wide TDM-specific exception to copyright infringement, allowing unrestricted use of copyrighted material with no clear option for rights holders to opt-out⁴. After receiving a tidal wave of backlash from the artistic community, the proposals in the new consultation reflect a more balanced perspective. The updated proposal for a TDM exception allows developers to scrape copyright-protected work for commercial purposes, insofar as they have lawful access and the rightsholder has not reserved its rights or ‘opted-out’⁵. This new approach aligns closely with the regulations introduced in the European Union’s Digital Single Market Copyright Directive in 2019.
A Balancing Act
Contrary to trademark and patent protections, there is no registration system for copyright in the UK. Copyright protections arise automatically on the creation of a work and typically expire 70 years after the death of the author⁶. Their validity is often only challenged once an infringement claim has been brought, and is by no means a straightforward test. On this basis, it can be argued that it is unreasonable to burden those creating AI tools (who are unlikely to be experts in copyright law) with the responsibility of deciphering what can and cannot be used in training, especially when mistakes could lead to severe legal consequences.
As above, the expansion and development of AI fundamentally involves data - and lots of it. If copyright protections are currently making developers reticent about using data that would otherwise be available to them, then this in turn halts the very development of technology which the government is looking to promote and incentivise.
Moreover, AI models often mimic rather than simply copy⁷, raising questions about whether they truly replicate original works. With most deep learning AI models, the specific training methods mean there is little, if any, transparency as to why they produce a given output. This lack of transparency creates a significant hurdle for rights holders trying to prove their work has been used without permission. In the recent High Court case, Getty Images v Stability AI, Mrs Justice Smith DBE contended the sheer number of data used to train the AI model in question, meant that trying to determine which works within the dataset were copyright protected ‘would be wholly disproportionate and practically impossible without significant resources’⁸. So, even if IP rights holders can opt-out, it is unclear whether they will have clear avenues to take action against developers who continue to infringe on their rights.
Perhaps new legislation would foster transparency and mutual respect between rights holders and developers, allowing all to work toward developing a balanced regulatory space, despite potentially competing interests. In the consultation, the government outlines that underpinning these new laws with a focus on greater transparency within AI training is key to making a new TDM exception workable. However, on a practical level the operation of this TDM will require clear guidance beyond the hopes of ‘increased transparency.’ Issues have cropped up at EU level with providing an opt-out mechanism, as there is no harmonisation as to which methods can be used in order to assert IP rights, beyond the fact that it must be done ‘in an appropriate manner’⁹ (although the directive does provide a few examples of what this could entail). There is a tangible risk that in searching for increased regulation, new legislation could serve only to inject greater legal uncertainty into an already unpredictable area. As Adam Buick, Lecturer at Ulster University highlights, promoting transparency is beneficial, but it alone cannot be enough to fundamentally address the ways in which legal frameworks must adapt to AI¹⁰.
Policymakers are sure to receive a wide variety of responses to this initial consolation, from stakeholders across technological and creative industries. The consultation has already prompted a letter to the Times opposing the changes, signed by prominent figures across the arts including Andrew Lloyd Webber, Ed Sheeran, Elton John, and other household names¹¹. UK creative industries have launched the ‘Make it Fair’ campaign in response to the consultation to protest the ‘existential threat’ that AI models pose¹². As the government navigates the competing interests of the art and tech sectors, it will be crucial to balance innovation and regulation - ensuring AI can evolve, while protecting the sanctity of IP regimes. Nevertheless, as AI becomes increasingly omnipresent, new regulations and legislation surrounding copyright will provide increased legal certainty for any individual or organisation which looks to utilise these tools.
The Role of Cambridge Mind Technologies
As AI development continues to develop, it is important to balance ethical data usage with technological innovation. Trust in AI-driven solutions depends on transparency—ensuring models are explainable, accountable, and built on fair data practices. At Cambridge Mind Technologies, we are strongly committed to exploring how AI can be developed responsibly, ensuring compliance with copyright laws while still driving forward groundbreaking advancements in conversational AI. By staying actively involved in emerging discussions on AI ethics, regulation, and copyright, Cambridge Mind Technologies strives to contribute to a future where AI technology remains both innovative and ethically sound.
With the developments in copyright legislation in mind, Cami has migrated to Llama, an open source AI model, with publicly available code, which provides increased transparency. This shift gives us greater control over training, response accuracy, and adaptability. By allowing us to further refine language use, this model ensures the AI communicates in a way that aligns with psychological best practices. With this platform, Cambridge Mind Technologies delivers a finely tuned product, harnessing the power of the LLMs while mitigating potential risk.
What are your thoughts on this? Please feel free to email hello@cambridgemindtechnologies with any opinions you have on this topic, we’d love to hear from you!
Image Source
References
Clarke, L. (2022) “When AI can make art – what does it mean for creativity?,” The Guardian , 12 November. Available at: https://www.theguardian.com/technology/2022/nov/12/when-ai-can-make-art-what-does-it-mean-for-creativity-dall-e-midjourney (Accessed: February 25, 2025)
Secretary of State Science, I. and T. (2025) AI Opportunities Action Plan. Available at: https://www.gov.uk/government/publications/ai-opportunities-action-plan/ai-opportunities-action-plan (Accessed: February 24, 2025).
Secretary of State Science, I. and T. (2024).Copyright and AI: Consultation. Available at: http://www.gov.uk/government/consultations/copyright-and-artificial-intelligence/copyright-and-artificial-intelligence#contents (Accessed: February 25, 2025).
O’Sullivan, N. et al. (2024) UK Government consultation on copyright and AI: A “win-win”?, Mishcon de Reya. Available at: https://www.mishcon.com/news/uk-government-consultation-on-copyright-and-ai-a-win-win (Accessed: February 24, 2025).
Shaw, J. and Fairhurst, O. (2024) UK Government Consultation: Copyright, AI and the Proposed “Opt-Out Model,” Lewis Silkin Insights . Available at: https://www.lewissilkin.com/insights/2024/12/17/uk-government-consultation-copyright-ai-and-the-proposed-opt-out-model-102jrlu (Accessed: February 24, 2025).
Copyright, Designs and Patents Act 1988 section 12(2). Available at: https://www.legislation.gov.uk/ukpga/1988/48/section/12
Clarke, L. (2022)
Getty Images and others v Stability AI [2025] EWHC 38 (Ch) [79]
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 and amending Directives 96/9/EC and 2001/29/EC, para 18
Buick, A. (2024) “Copyright and AI training data—transparency to the rescue?,” Journal Of Intellectual Property Law and Practice [Preprint]. Available at: https://doi.org/10.1093/jiplp/jpae102.
Sellman, M. and Smyth, C. (2025) “Don’t let AI steal our copyright, giants of the arts tell Labour,” The Times, 24 February. Available at: https://www.thetimes.com/uk/technology-uk/article/dont-let-ai-steal-our-copyright-giants-of-the-arts-tell-labour-latest-news-rfw55l20 (Accessed: February 25, 2025)
Thomas, D. and Gross, A. (2025) “Creative industries protest against UK plan about AI and copyright,” Financial Times, 24 February. Available at: https://www.ft.com/content/38585a82-cffd-4144-9969-82e94cbb2168 (Accessed: February 25, 2025).
Department for Science, Innovation and Technology. (2025). Prime Minister sets out blueprint to turbocharge AI. [Press release]. GOV.UK. Available at: https://www.gov.uk/government/news/prime-minister-sets-out-blueprint-to-turbocharge-ai (Accessed: 26 February 2025).
Author: Charlotte Westwood, Cambridge Mind Technologies Volunteer
Charlotte is a First Class Law graduate from the University of Cambridge and currently works as a paralegal in a boutique firm that specialises in advising families and individuals. She is passionate about AI ethics and interested in how the growing prevalence of AI will impact both our legal systems and personal lives. Charlotte volunteers at Cambridge Mind Technologies because she is inspired by how the project is harnessing innovative technology and technical expertise, to create a tool which will provide mental health support to young people as they navigate troubling times. Through contributions to the blog, she aims to shed light on the potential legal reforms regarding AI and their impact on start-ups, showcasing how Cambridge Mind Technologies is uniquely positioned to thrive in this rapidly evolving space.
