UKIPO Launches New Consultation on Copyright and AI Interplay
In Short
The Situation: The UK Intellectual Property Office ("UKIPO"), in collaboration with the UK Department of Culture, Media and Sport ("DCMS") and the UK Department for Science, Innovation and Technology ("DSIT"), has initiated a Consultation on Copyright and Artificial Intelligence.
The Aim: The consultation seeks to address the interplay between copyright law and artificial intelligence ("AI"). Presently, the application of UK copyright law to the training of AI models remains ambiguous, presenting potential challenges for both AI developers and rights holders.
Looking Ahead: The consultation period will last 10 weeks, concluding on 25 February 2025. The outcome could lead to significant changes in UK copyright law, aligning it more closely with recent developments in the European Union ("EU"), as the United Kingdom aims to strengthen its position in the AI sector.
Balancing Data Access and Copyright in AI Model Training
The training of AI models necessitates the use of vast quantities of data. This data can come from a number of sources, including publicly available data, internal data from within an organisation and scraping internet data. Internet scraping involves the mass extraction of information from public websites, which is then exported into structured formats for use in training AI models. However, the practice of scraping data can be non-discerning as between material that is copyright protected, and that which is not. The scraping of copyright protected works without the consent of rights holders has sparked significant debate and raised questions of intellectual property infringement before policymakers and courts around the world.
The UK government appears to be following earlier consultation attempts and is keen to explore the possibility of a copyright exception to support AI developers by ensuring legal access to high-quality data, while also enabling rights holders to obtain and control remuneration. The proposed approach is broadly akin to Article 4 of the EU Digital Single Market (DSM) Directive, which the United Kingdom previously opted not to implement.
The consultation generally seeks comments on three areas:
1. Expansion of the TDM exception. Currently the text and data mining ("TDM") exception in Section 29A of the Copyright, Designs and Patents Act 1988 ("CDPA") applies solely to non-commercial research. It is proposed that the exception be expanded to permit TDM for commercial purposes and apply by default. Notably, however, rights holders will have the ability to reserve their rights and opt out of the exception, prohibiting data mining of their materials. Instead, a licensing mechanism could be used, which would allow rights holders to seek remuneration for use of certain of their data.
The current consultation questions ask whether respondents agree with the proposed approach or if alternative options are preferred, including the option of no change at all.
Questions on this matter include:
- 'What influence, positive or negative, would the introduction of an exception along these lines have on you or your organisation? Please provide quantitative information where possible.'
- 'What should be the legal consequences if a reservation is ignored?'
2. Transparency measures. It is proposed that the TDM exception would be underpinned by stringent transparency provisions. Such measures could require AI developers to disclose the materials used for training and the content generated, and to label clearly AI outputs, with the aim of benefiting rights holders and the public. However, the practical challenges surrounding these measures could be significant given the large quantities of data involved and legitimate reasons for rights holders to limit such information (e.g., trade secrets). For this reason, input is sought on how best to implement transparency measures in a balanced and proportionate manner.
Questions on this matter include:
- 'Do you agree that AI developers should disclose the sources of their training material?'
- ' If so, what level of granularity is sufficient and necessary for AI firms when providing transparency over the inputs to generative models?'
- 'Where possible, please indicate what you anticipate the costs of introducing transparency measures on AI developers would be.'
3. Removing protection for the outputs of generative AI. Section 9(3) CDPA is also being re-examined. The section outlines that computer-generated works ("CGW") without a human author are deemed to be owned by the person who made 'the arrangements necessary for the creation of the work to be undertaken.' This exception has faced significant criticism, and comments are sought on the proposal to remove it.
Questions on this matter include:
- 'Are you in favour of removing copyright protection for computer-generated works without a human author?'
- 'Would the removal of the current CGW provision affect you or your organisation? Please provide quantitative information where possible.'
Four Key Takeaways
- The consultation period is scheduled to last for 10 weeks, concluding on 25 February 2025, and covers numerous sub-issues and surrounding topics to the three mentioned above.
- It is likely that robust responses will be submitted from both copyright owners (particularly those in the creative industries) and those behind leading AI models. Indeed, this is not the first time that the United Kingdom has consulted on the amendment of section 29A. Toward the end of 2021, the UKIPO floated a proposal to extend the TDM exception, albeit without an opt-out provision for rights holders. This proposal was ultimately withdrawn following input from academic experts and industry groups, where the committee determined that the proposal insufficiently addressed the potential harm to the creative industries.
- If the proposals suggested in the consultation come to fruition, this would mark a drastic shift in UK copyright law, albeit in the direction of travel already seen in the EU.
- Legislative and policy change around AI and copyright has been advancing rapidly abroad, particularly in the EU, while the United Kingdom's stance has, to date, remained unchanged. As the United Kingdom seeks to position itself as a key competitor for AI, it is unsurprising that these questions are being revisited, and the momentum for change has increased.
