The consultation recognises the importance of ensuring that existing ambiguity in the law is removed and ensuring that the UK’s copyright laws are internationally interoperable. 

In line with the EU, the consultation proposes implementing a new provision which would permit text and data mining of copyrighted works for AI development (including for commercial purposes) provided that AI developers have lawful access. Such exception would encompass works available online or under contractual agreements (for example via a subscription) but would not apply where a rights holder has explicitly reserved their right to prevent use of their works for training AI. 

The consultation welcomes the development of technical solutions to restrict the use of copyrighted works AI training where the rights holder has enforced a rights reservation. However, it notes that a lack of standardisation in approaches is challenging for rights holders to navigate and that standardisation would enable rights holders to signal their preferences regarding the use of copyrighted works and AI developers to respect such preferences.

The consultation underscores the importance of fair and transparent licensing agreements for AI training data, particularly for individual creators. It proposes collective licensing and data aggregation services as potential solutions for streamlining the licensing process to ensure that rights holders can seek fair remuneration for their work, and that AI developers have viable means to access datasets.

The consultation proposes mandating transparency from AI developers regarding their training data, which may include disclosing the use of specific works and datasets, details of web crawlers (ownership, purpose, etc.), and potentially requiring records and compliance demonstrations. This would inform rights holders about their work's usage but may pose practical challenges, especially for smaller firms, and those concerned about the impact on commercial confidentiality.

The consultation seeks information about how to tailor requirements based on company size and specific application of AI technologies, aiming to ensure that any new regulations are proportionate and do not stifle AI research and innovation. 

Currently, the Copyright Designs and Patents Act 1988, provides copyright protection to certain types of works entirely generated by a computer. The consultation seeks input on the legal clarity, effectiveness, impact, and moral argument of this protection in the age of advanced AI, seeking views on whether to retain, clarify, or potentially remove it. 

The consultation reiterates that content generated by an AI model will infringe copyright law if it reproduces a “substantial part” of a protected work. The liability of both AI developers and users in such cases is also considered. The consultation seeks views on whether there are specific areas where copyright law may be deficient or where they may be barriers to enforcement. 

The consultation proposes regulating and mandating consistent labelling of AI-generated content to better inform users of how content was created.  

The government recognises concerns about AI being used to create realistic but fake images, videos, or audio recordings of people ("deepfakes"). The consultation highlights a complex legal landscape, the need to balance free speech with protecting individual rights, and seeks input on whether current laws are enough to deal with this.