Artificial intelligence (AI) is transforming how content is created, accessed, and monetised in the United Kingdom. Modern AI systems, particularly large-scale models, depend on vast quantities of data for training, refinement, and deployment. Much of this data is protected by copyright, placing AI development in direct interaction with a legal framework that was not designed with such technologies in mind.
As a result, UK copyright law has become a central point of tension between technology developers and the creative industries. While the Copyright, Designs and Patents Act 1988 (CDPA) continues to govern these issues, its application to AI raises complex and, in many cases, unresolved legal questions. These arise not only at the point of output, but throughout the entire AI lifecycle, from data acquisition to real-time system use.
This article examines the UK copyright framework for AI, how it operates in practice, the legal status of AI-generated works, and the evolving policy landscape shaping future reform.
The UK Copyright Framework Across the AI Lifecycle
UK copyright law protects a broad spectrum of works, including literary, artistic, musical, and dramatic works, alongside entrepreneurial categories such as sound recordings, films, broadcasts, and databases.
These categories reflect a distinction between authorial works, which require originality in the form of human intellectual creation, and entrepreneurial works, which protect the investment involved in their creation rather than creativity. This distinction is particularly significant in the context of AI, where the role of human input is often unclear or diluted.
Copyright law applies at every stage of the AI lifecycle:
- Data acquisition: developers compile large datasets, often containing protected material.
- Training: Protected works are copied and analysed, engaging the reproduction right.
- Fine-tuning: Additional layers of copyrighted content are introduced, particularly in commercial AI applications.
- After deployment: Systems that retrieve and process external content in real time may generate transient copies or outputs may reproduce protected works in a manner that gives rise to infringement.
This continuous engagement with copyright underscores the scale of the challenge: AI does not simply interact with copyright law, it operates within it at every stage.
Copyright Exceptions in the UK: Scope and Limitations
The UK adopts a narrow and rules-based approach to copyright exceptions. Unlike jurisdictions that rely on flexible doctrines such as “fair use,” the UK framework is based on specific statutory exceptions with defined boundaries.
The Text and Data Mining (TDM) Exception
The TDM exception allows copying for computational analysis but is limited to non-commercial research and subject to conditions such as lawful access and acknowledgement. In practice, this means that most commercial AI development falls outside its scope.
The Temporary Copying Exception
This exception provides limited flexibility by allowing transient or incidental copies that are integral to a technological process and without independent economic significance. This applies to activities such as caching, buffering, and real-time inference, where copies exist only momentarily.
However, it does not apply to stored or reusable copies, training datasets, or copies with economic value. As a result, it offers only partial accommodation for AI technologies and does not resolve the broader legal challenges associated with large-scale data use.
Taken together, these limitations mean that lawful AI development in the UK typically depends on obtaining licences from copyright holders.
How the System Operates in Practice
In practice, the UK copyright framework for AI operates through a combination of licensing, negotiation, and evolving market practices rather than clear judicial authority.
Licensing and Market Practice
Licensing has emerged as the primary mechanism through which AI developers access copyrighted material, with a growing number of commercial agreements being concluded across different sectors.
However, these arrangements are often opaque and unevenly distributed. Not all right holders are equally positioned to negotiate, and the absence of standardised frameworks can lead to fragmentation and inefficiency.
Legal Uncertainty and Enforcement Challenges
At the same time, legal uncertainty remains a defining feature. There is no settled consensus on whether certain forms of AI training constitute infringement, how exceptions should be interpreted across the lifecycle, or where liability ultimately lies in relation to outputs.
Enforcement presents further challenges. Without transparency around training datasets, right holders face significant barriers in identifying and proving infringement. Even where infringing outputs are identified, attributing responsibility, whether to users, developers, or distributors, can be difficult.
The result is a system that is operational but unsettled, characterised by private ordering and ongoing negotiation rather than legal certainty.
Ownership of AI-Generated Content Under UK Law
Ownership is a central issue in the relationship between AI and copyright law.
Human-Authored Works Using AI
Where AI is used as a tool and a human exercises creative control, the resulting work will generally qualify for copyright protection as an original work. The key requirement remains human intellectual creation.
Computer-Generated Works
More complex issues arise where works are generated without human authorship. UK law recognises computer-generated works, assigning authorship to the person making the necessary arrangements for their creation and granting a fixed term of protection.
However, this framework is increasingly questioned in the context of modern AI, particularly where content is generated at scale, and human creativity is minimal or absent. This raises concerns about whether such protection continues to serve its original purpose of incentivising creativity.
In parallel, certain AI outputs may attract protection as entrepreneurial works, reflecting the investment involved in their production rather than any creative input.
At the same time, AI-generated outputs may themselves infringe existing copyright where they reproduce a substantial part of protected works. Liability in such cases may extend across multiple actors, including users, developers, and distributors, and may also arise indirectly through the use or importation of infringing systems.
UK Policy Developments on AI and Copyright
The UK policy debate on AI and copyright has focused on four main approaches: maintaining the current framework, strengthening copyright protections, expanding data mining exceptions, or introducing a hybrid model combining licensing, exceptions, and transparency.
Shift Toward a Balanced Approach
In December, the policy trajectory appeared to favour expanding access to copyrighted material through broader exceptions, with the aim of supporting AI development.
However, this position has since shifted following strong opposition from the creative industries highlighted concerns about uncompensated use and the potential erosion of existing markets. There was also recognition that expanding exceptions would not address underlying issues such as transparency and enforceability.
Current Direction of Reform
The current direction of travel reflects a more balanced approach. Licensing is expected to remain central, but with increasing emphasis on improving how those markets function in practice. Transparency has emerged as a key priority, seen as essential to enabling enforcement and ensuring fair remuneration.
At the same time, there is growing support for targeted intervention, particularly in areas such as clarifying liability, supporting licensing frameworks, and reassessing the role of computer-generated works within the copyright system.
Conclusion
The UK copyright framework is under increasing pressure as it adapts to AI. As it continues to apply across the AI lifecycle, its limitations are increasingly exposed by the scale and complexity of modern AI systems. In practice, the system is sustained by licensing and market mechanisms, but characterised by uncertainty and limited transparency. Policy has moved away from broad exceptions toward a more nuanced approach that seeks to balance innovation with the protection of creative industries.
The direction of reform is therefore likely to be incremental rather than radical, focusing on improving transparency, strengthening market mechanisms, and clarifying key areas of law. The challenge for policymakers will be to ensure that the framework remains both adaptable to technological change and capable of sustaining the economic foundations of the UK’s creative sector.
How To Get In Contact
To find out more or if you require assistance with these matters, speak with our Intellectual Property Team on +44 (0)204 600 9907 or email info@culbertellis.com.
Accurate at the time of writing. This information is provided for general information purposes only and should not be relied upon as legal advice.
.png)
