UK Seeks Clarity on Copyright for AI Training

The United Kingdom is considering a significant change to its copyright law in response to the rising influence of artificial intelligence (AI) firms. The government has proposed an exception to the law that would allow AI companies to use copyright-protected material for training purposes in a commercial context.

This proposal was unveiled during a 10-week consultation that began on December 17. The goal of this initiative is to provide clarity regarding how copyrighted materials can be utilized in the development of AI models. The government hopes that by doing so, it can foster growth in both the AI sector and the creative industries.

According to various UK government departments, including the Intellectual Property Office, the Department for Science, Innovation and Technology, and the Department for Culture, Media and Sport, this consultation aims to improve trust and transparency between AI developers and copyright holders. The focus will be on how copyrighted material can be licensed, ensuring compensation for data owners, and enhancing access to quality data for AI developers.

The government acknowledges that current uncertainties surrounding copyright law are hindering both creative and technological sectors. Data owners face challenges in receiving payments for the use of their work, while AI developers navigate legal risks concerning the material they utilize.

To mitigate these issues, the proposed exception would allow AI training on copyrighted material while still permitting copyright holders to maintain their rights. This change aims to provide data owners with greater certainty and control, making it easier for them to negotiate licensing agreements with AI firms. Simultaneously, AI developers would gain clarity about what content they can use without facing legal repercussions.

Additionally, the consultation seeks to implement new requirements for AI developers to enhance transparency regarding the content used for model training.

However, this initiative has not received unanimous support. Some book publishers and news organizations have criticized the proposal. Dan Conway, CEO of the Publishers Association, expressed concerns that the suggested measures lack testing and evidence. He argued that there has not been a compelling case made for a new copyright exception.

Owen Meredith, chief executive of the News Media Association, voiced similar sentiments, asserting that the government’s approach does not adequately address the core issues. He emphasized that news publishers should retain control over their content and receive fair compensation for its use. Instead of introducing complex systems such as rights reservations, he urged the government to focus on enhancing transparency within the existing copyright framework.

Adding to the debate, Gill Dennis, a senior lawyer with Pinsent Masons, highlighted the controversy surrounding the government's proposed opt-out approach for content creators, emphasizing the significant technical challenges it poses in practice.

A prominent example of the complexities surrounding copyright and AI training is related to major AI innovators like OpenAI. The company has been criticized for its unclear policies regarding the data used for training its models. Although it utilizes various copyrighted materials, its large language models do not directly reproduce the data, complicating the ability for copyright holders to demonstrate infringement.

Recent legal actions illustrate the rising tension between news publishers and AI firms. In a notable case, several Canadian news publishers filed a lawsuit against OpenAI, seeking substantial damages for copyright infringement. Meanwhile, the New York Times is reportedly gathering evidence for its own lawsuit, claiming that ChatGPT has been trained on a vast number of articles produced by the publication.

UK, copyright, AI