The UK Government’s consultation on Copyright and AI

The question

How is the UK Government considering striking the balance between the protection of copyright in materials and the training of AI?

The key takeaway

Companies, particularly those in creative industries, should consider the Government’s ongoing consultation to shape the treatment of copyrighted materials and training of AI models, specifically on control, access and transparency. Companies should take the opportunity to feed into the consultation to shape the protections for their works. Similarly, developers of AI models should respond to help influence how AI models may be lawfully trained in the future.

The background

The Government began a consultation on copyright and AI in response to calls for a more certain, transparent and commercially attractive approach to use of copyrighted data. This was against the backdrop of increasing interest from AI developers to maximise access to materials to train AI models. As a result, owners of copyrighted materials have expressed their reservation about the practically unfettered use of their content in training datasets. The consultation lists out a variety of options for the Government to pursue which range from doing nothing to creating certain exemptions from copyright in the context of training an AI model.

The development

The Government aims to focus its reform on:

• enhancing rights holders’ control over their material and renumeration
• supporting access to high quality materials, bolstering the UK as a competitive market for AI developers to train their models, and
• achieving greater transparency in the training process to build trust with creators and consumers alike.

Companies should take note of the Government’s headline proposed approach which is to introduce a new text and data mining exception to copyright laws. This carve out will allow copyrighted content to be used for AI training, unless creators have opted out. There remains debate about what a standard “opt out” process may look like. For example, copyright owners may be able to: (i) embed metadata instructions on the copyrighted material stating that it cannot be used for AI training purposes; (ii) employ the “robots.txt” standard to prevent trawling; and (iii) and/or register on “Do-Not-Train” sites.

Additionally, the Government aims to bolster transparency requirements for the benefit of both creators, who will want to know when their works are used for training purposes, and consumers, who will want to know the provenance of the content they may be viewing. This may require explaining how AI models have acquired data and flagging the content generated from it.

Why is this important?

The impact of the consultation will be wide reaching across the creative and tech sectors in the UK. Owners of copyrighted materials will need to consider the level of technical and practical difficulty in effecting an opt-out as set out in the consultation. AI developers will also need to ensure that their models will be trained lawfully under the new proposed approach and to protect their investment in the training process.

Any practical tips?

Copyright holders should review their materials and assess the practical steps needed to protect their content. Developers of AI must also consider their dataset sources and training practices, and how they may be affected by the proposed approach of the Government. These considerations should shape both sets of stakeholders’ contributions to the Government’s consultation.

Spring 2025