IBIL Responds to Government Consultation on Copyright and AI

Last week, the UCL Institute of Brand and Innovation Law (IBIL) submitted its response to the UK Government’s latest Consultation on AI and Copyright. The IBIL Response was drafted primarily by Dr Alina Trapova. Dr Trapova's research interests focus on copyright law and the implications of machine learning and artificial intelligence on the creative industries.

The ongoing battle between the creative industries and AI developers has been widely reported in the mainstream press. While the creative industries rely upon copyright protection to generate revenue from their work, AI companies have already made widespread use of copyright material for training AI models without always (often) securing a licence to do so. This has resulted in high profile litigation in numerous jurisdictions, as well as new regulation in the form of hard and soft law. But until a substantive court judgment has been handed down, uncertainty persists on many fronts. Do AI training activities involve unlawful copying? If so, what (if any) copyright exception applies? What is the legal status of models trained outside of the UK? How do right holders reserve their rights successfully if they don't want their works being used for AI training? What transparency obligations must be introduced for AI companies?

Since 2014, UK copyright law has included a text and data mining (TDM) exception, which could extend to AI training, but this provision provides a defence covering non-commercial research activities only. Previous attempts to introduce a far broader TDM exception covering any commercial use (including AI training) were withdrawn following an outcry from the creative industries. Subsequent plans to introduce a voluntary code to regulate data-scraping also failed because no consensus could be reached - an impasse that lead to the current consultation.

The UK Government's Consultation, launched in December 2024, sought views on how best to balance the competing aims of the growing the UK’s AI sector without detriment to the country’s successful creative industries. The Consultation set out three main objectives:

1. Supporting right holders’ control of their content and ability to be remunerated for its use;
2. Supporting the development of world-leading AI models in the UK by ensuring wide and lawful access to high-quality data; and
3. Promoting greater trust and transparency between the sectors.

The Consultation document sought endorsement of the Government’s preferred solution - called 'Option 3' - modelled on the TDM exception recently introduced into EU copyright law. This was presented as the option most likely to met the objectives of control, access and transparency, outlined above. Option 3 proposed a data mining exception for commercial use which allows right holders to reserve their rights (‘opt-out’), underpinned by supporting measures on transparency. The Consultation document suggested that once a clearer copyright framework was put in place for TDM activities, it was more likely that right holders’ reservations of rights would be respected, resulting in increased licensing of these works, in turn leading to higher revenues to right holders.

The IBIL Response focussed on four main points.

Firstly, as a preliminary point, the Response stressed that any changes in copyright law should focus on safeguarding the interests of ‘primary creators’ in recognition of the fact that creators' interests do not always coincide with those of the right holders that control the commercialisation of protected works. Therefore, remuneration of primary creators should feature strongly in whatever new forms of laws are being proposed.

Secondly, although the Government’s preferred Option 3 might be the best of the options proposed in the Consultation, further revision would be required in order for the new exception to meet the stated objectives. As Option 3 seeks to align the UK with the recent EU exception, the UK should learn from the EU experience thus far, and clarify the meaning of ‘lawful access’. Additionally, success of this option would be contingent upon careful thought about the rights reservation mechanism as well as and the requirements for transparency.

Next, with regard to the rights reservation mechanism, the Consultation document indicates that all copyright works which are made available online would need to include a machine-readable mechanism that enabled systems engaged in data mining to easily identify works that may be mined lawfully. The Response highlights the need to identify a robust and standardised protocols for this which, while needing to be accurate, does not place too high a burden on right holders. It also highlights short-comings with robot.txt and use of metadata – which the Consultation document flagged as potentially suitable 'opt-out' tools.

Finally, in terms of transparency on the part of AI developers, the Response urges the UK to align itself with the work being undertaken at the EU level. Having different transparency regimes in place in the UK and EU would be particularly detrimental to AI start-ups, micro businesses and SMEs, who would be likely to incur significant compliance costs which might prove too high a barrier to entry.

On all of these important points, the Devil will be certainly be in the detail!

Additionally, the Response critiqued the shortcomings of the current UK TDM exception in s29A of Copyright Designs and Patents Act 1988 (CDPA), emphasising the need to remove the copying prohibition aspect of the provision, as well as the need to clarify the ‘lawful access’ terminology. It also endorsed the Consultation's proposal to repeal s. 9(3) of CDPA, the provision which currently affords copyright protection to computer-generated works. Here, the Response argued that the current provision is flawed with uncertainty and removing protection for such outputs will only have a positive economic impact.

Find out more about the consultation here.

Read the IBIL Response here.