The UK government's open consultation on Copyright and Artificial Intelligence, published on 17 December 2024, addresses the complex interplay between copyright law and the burgeoning field of artificial intelligence (AI). The consultation seeks views on creating a framework that balances the interests of right holders and AI developers, fostering innovation while protecting human creativity, and acknowledges the current legal uncertainties surrounding the use of copyright works in AI training and the need for a balanced approach that supports both sectors.

Key objectives

The consultation outlines three primary objectives:

  • Supporting right holders' control and remuneration: Ensuring that creators can control the use of their works and receive fair compensation.
  • Supporting AI development: Providing AI developers with lawful access to high-quality data to train world-leading models in the UK.
  • Promoting trust and transparency: Enhancing transparency between right holders and AI developers to build trust and ensure compliance with copyright law.

Proposed approaches to using copyright works for training AI models

The consultation explains that leaving current laws unchanged would maintain legal uncertainty and difficulties for right holders and AI developers, continuing to make it difficult for right holders to seek remuneration and enforce their rights, while AI developers would face ongoing legal risks. The government does not wish to prolong this legal uncertainty, as it does not meet its objectives.

The consultation outlines several approaches the government has considered to clarify copyright law and meet its objectives:

Strengthening copyright

One option is to require explicit licensing for all uses of copyright works in AI training. This approach would provide clear legal certainty and ensure that right holders are compensated. However, it could make the UK less competitive compared to jurisdictions with more permissive laws, potentially driving AI development elsewhere, which is not aligned with the government's "AI Opportunities Action Plan" published on 13 January 2025, which states the government must position the UK to be an AI maker, not an AI taker.

Broad data mining exception

Another option is to introduce a broad data mining exception, allowing AI developers to use copyright works without permission. While this would boost AI innovation, it would undermine the rights of creators and could conflict with international copyright treaties.

Data mining exception with rights reservation

The government's preferred approach is a data mining exception that allows right holders to reserve their rights. This would enable AI developers to use works unless the right holder has explicitly opted out. This approach aims to balance access and control, supported by transparency measures to ensure compliance.

Finding a solution is vital if the government wants the UK to become a hub for AI and encourage data training within its borders. Without an appropriate text and data mining exemption for information in the public domain (that cannot be circumvented by contractual restrictions), AI developers will likely seek more permissive jurisdictions, undermining the UK's competitiveness in the market. The government seems hopeful that its preferred approach will dissuade AI model operators from choosing other jurisdictions to build and train their models, while giving right holders confidence that there are suitable transparency and remuneration requirements and mechanisms in place to license works to train AI models.

However, it will remain a challenge for right holders to prove that AI outputs trained abroad and imported into the UK infringe UK copyright law when the software is run and trained outside of the UK (although we are keeping an eye on the Getty Images v Stability AI case where these points are being raised). While right holders may be opposed to their works being used to train AI models in such a way, this approach may be adopted by AI developers to circumvent copyright laws (and, if the preferred approach is adopted, to avoid paying remuneration), making legal action costly and judgments more difficult to enforce.

The consultation states the government is considering a direct intervention by legislating to clarify the position on copyright infringement (rather than waiting for a resolution through ongoing legal cases) but it does not set out if it will seek to impose extra-territorial copyright laws in respect of AI outputs imported into the UK where the AI model has been trained by works benefiting from UK copyright protection. A similar problem arises where seeking to impose transparency requirements on AI models; will the government seek to apply transparency requirements in an extra-territorial manner where the AI output is imported into the UK?

By embracing an AI training exception, the UK can legitimately encourage AI models to be developed and trained in the UK and meaningfully implement the "AI Opportunities Action Plan". The results of the consultation are also expected to inform the government's policy and it is worth noting recommendation 24 in the Action Plan: "Reform the UK text and data mining regime so that it is at least as competitive as the EU"; the question is: how far will the UK government go to attract AI businesses and will this be a boon or detriment to right holders?

Transparency and technical standards

The consultation emphasises the need for greater transparency from AI developers about the sources of their training data and highlights the importance of standardising rights reservation protocols to make it easier for right holders to control the use of their works.

The expectation is set that suppliers use AI models transparently, ensuring that the sources of their training data are traceable and disclosable. This transparency is crucial for defending against allegations and claims related to the use of AI, which may be subject to regulation, such as anti-discrimination and data protection laws.

There is a growing trend of requiring transparency and disclosure of information in contracts where AI is involved in the services and goods being provided. By introducing legislation to ensure transparency for AI model use in the UK and the adoption of appropriate standards, the UK could build trust and ensure legal and ethical compliance, further supporting the objective to promote AI development in the UK.

AI outputs and copyright

The consultation addresses the copyright status of AI-generated outputs and proposes that outputs incorporating substantial parts of copyright works should require a licence, while also considering the protection of purely computer-generated works.

If a text and data mining exception is to be adopted, the government could consider preventing copyright protection from applying to AI-generated works, giving some comfort to right holders that human intellectual endeavour is protected by copyright, while AI-generated works are not. However, an issue arises where a user puts in a prompt to make changes that are easily attributable as the author's own intellectual creation, such as a clear instruction to replace a specific word with another. Should the author then not benefit from copyright protection of the AI output as there is sufficient proximity between the intention and understanding of what the AI output will be? The courts will likely be grappling with this issue over the next few years.

One resolution could be to allow copyright to subsist in an AI output only where the prompt user would be the sole author of the copyright had they asked another human to create the work with the same instructions and they produced the same as the AI output. This adopts a similar approach taken in the US copyright system, where AI-generated elements cannot be copyright protected and need to be identified when seeking to register copyright.

The consultation closes at 23:59 on 25 February 2025. Responses may be submitted via Citizen Space or via email to: copyrightconsultation@ipo.gov.uk.

If you would like any assistance in preparing your response to the consultation, please contact our Intellectual Property team.

This article is for general information only and reflects the position at the date of publication. It does not constitute legal advice.