The U.K. government is testing a new approach to copyright and AI training: an opt-out model that would require rights holders to take active steps if they do not want their intellectual property used to train AI systems.
The consultation places one of the most contested questions in generative AI at the center of policy debate. AI developers want broad access to high-quality data, while writers, artists, musicians, filmmakers, game creators, and other rights holders want control, permission, and payment when their work is used.
What the U.K. is considering
The proposed framework would create a mechanism for right holders to reserve their rights. If they do so, they could license their work and be paid when it is used in AI training.
Alongside that, the government is considering an exception that would support large-scale use of a wide range of material by AI developers where rights have not been reserved. In plain terms, the default could shift toward allowing AI training unless a rights holder has opted out.
The government says its aim is to support both the creative industries and the AI sector. It presents the consultation as a way to build a copyright and AI framework that rewards human creativity, encourages innovation, and gives both sectors more legal certainty.
Future policymaking could include legislation “to provide legal certainty,” although the government has not yet decided whether it will take that route.
Why AI training has become a copyright flashpoint
Generative AI models are trained on vast quantities of data. That has pushed intellectual property concerns into the open, especially because many creatives argue their work is being processed without permission or compensation.
The concern is not only that creative work may be used in training. It is also that AI tools can generate competing output across text, visuals, audio, or combinations of those formats.
The source article identifies several creative fields as viable targets for generative AI, including:
- visual arts
- music
- film production
- video games
These are areas where skilled human production processes can be challenged by scalable AI tools. Those tools use prompts to generate output based on statistical analysis of patterns in training data.
Large language models such as OpenAI’s GPT, which underpins ChatGPT, have made AI training practices much more visible. The period when AI startups could quietly scrape the web for free training data without attracting much attention is over.
Some AI companies are striking deals to license certain types of content for training data. At the same time, a growing number of lawsuits are challenging the unlicensed use of intellectual property for AI training.
The burden on rights holders
The central controversy is where the burden should fall. Under an opt-out regime, rights holders would need to act if they wanted to keep their work out of AI training datasets.
That structure could favor organizations with the resources to monitor, reserve, and enforce their rights. Smaller creatives may face a harder task. They may have less capacity to track how their work is used, understand new technical systems, or negotiate licensing terms.
For that reason, the approach is unlikely to be universally popular with the creative sector. The source article notes that AI companies have been actively lobbying for an opt-out regime, which helps explain why some creatives may view the proposal as tilted toward AI developers.
The government frames the proposal as a balance. Rights holders would retain a route to control and payment when they reserve their rights. AI developers would gain a clearer legal basis to train models in the U.K. using copyright material where rights have not been reserved.
That balance is difficult. The government’s stated goals include supporting rights holders’ control of their content and ability to be remunerated, while also enabling the development of “world-leading AI models in the UK by ensuring wide and lawful access to high-quality data.” Those goals can pull in different directions.
Transparency is the condition that matters
The government says greater transparency from AI developers is a prerequisite for the proposed approach to work. Without it, rights holders may not know what material has been used, how it was obtained, or what content AI systems are producing.
The consultation specifically seeks views on transparency around three areas:
- the material AI developers use to train models
- how AI developers acquire that material
- the content generated by AI models
This matters because an opt-out model depends on practical visibility. If creators cannot understand whether and how their work is entering AI systems, the right to reserve rights may be difficult to exercise in practice.
The government also points to the need for simple technical means for creators to exercise their rights, either individually or collectively. It says AI companies and creative industries would need to work together to create new technical systems that support greater control and licensing of intellectual property.
That is one of the biggest implementation questions. A policy can state that creators have control, but the system will only work if creators can use that control without unreasonable complexity.
What happens next
The consultation runs for 10 weeks and closes on February 25, 2025. Web submissions can be made through an online survey.
The government describes the consultation as an opportunity for anyone interested in the issue to share views and provide evidence about the economic impact of the proposals. It has also committed to a program of wider engagement activity during the consultation period to ensure a full range of views is heard.
The outcome could shape how the U.K. positions itself on AI training, copyright, licensing, and creative rights. For AI developers, the key issue is whether they get a clear legal pathway to use large volumes of material. For rights holders, the question is whether control and remuneration are meaningful in practice, not just available in principle.
The proposal therefore turns on more than the legal default. It also depends on transparency, usable technical systems, and whether the final framework gives both sectors enough certainty to operate without downgrading one side’s interests for the benefit of the other.