There was otherwise strong support in the creative industries for statutory transparency measures in relation to AI model training and the works used in such training, while tech firms had mixed views and generally supported a light-touch approach.

In response to the consultation, the government formed expert working groups to develop transparency rules, technical standards, licensing frameworks and creator remuneration mechanisms.

The consultation feedback suggested that AI developers may be required in the future to obtain licences in relation to each copyright work they use to train their AI tools in the UK; however, as below, this is not something the government is planning to introduce at this time. Currently, AI developers may be able to rely on limited exceptions and grey areas within the existing regime, such as the text and data mining exception for non-commercial use under the Copyright, Designs and Patents Act 1988 (“CDPA 1988”), as well as implied licences, and other fair dealing exceptions to use data to train their tools. However, if such a mandatory licensing regime were to be implemented in the future, this would represent a shift to a strict rightsholder control model, which will favour the creative industry, but may deter AI developers from entering the UK market through significantly increased compliance costs.

We also speculated whether the government may favour implementing a collective licensing regime (in the same way as e.g. music copyright). Indeed, to demonstrate that licensing may be the answer, the CLA (the Copyright Licensing Agency) together with its member associations, PLS (Publishers’ Licensing Services) and ACLS (Authors’ Licensing and Collecting Society) announced in April 2025 that it would be developing such a collective licence for release in Q3 2025. No such licence has been released as yet, however.