During 2026, governments worldwide are consulting on possible new policies and legislation aimed at keeping pace with the impact of Generative Artificial Intelligence (GenAI). Consultations are intensifying, as authors and governments continue their dialogue to find solutions that strike a fair balance between protection of authors’ copyrights, and commercial industries’ unrestricted use of authors’ works using GenAI tools.
Over the past five years, use of AI has proliferated globally, and authors in all creative fields have one principal concern: that their original works are being used – without permission – to generate text, images, video, audio, software code, or other forms of data. In the UK, the Copyright, Designs and Patents Act 1988 (CDPA) deals with uses and abuses of digital technology. Although copyright may be violated using mechanical or digital means, it is unclear whether GenAI training without permission is ‘substantial’ enough copying to constitute copyright infringement; and, to date, no major court cases have settled the question.
For the past year, the UK Government has been consulting on its preference for changing copyright law to introduce an author’s so-called ‘opt-out’ mechanism for text and data mining (TDM) of copyright-protected works. Such a mechanism is a legal and technical process: it puts the onus on copyright owners to ‘reserve their rights’ and prohibit the automated analysis of content for training GenAI models. In other words, unless the author has explicitly ‘opted out’ by reserving their rights, GenAI developers would generally be permitted to mine text and data of works.
When authors make their works accessible online, an opt-out would need to be expressed in a ‘machine-readable’ format, so that GenAI crawlers can automatically recognise and respect the constraint. Opt-out restrictions typically apply only to commercial TDM, but not to scientific research conducted by non-profit or cultural heritage institutions. Authors fear that opting out places an unreasonably onerous burden on them to proactively declare their objection. Doing so would require authors to acquire technical digital skills for implementing opt-outs via a variety of common ‘machine-readable’ methods. ‘Robots.txt’ uses standard web crawler instructions to block specific AI bots. ‘AI.txt’ is an emerging protocol designed specifically for expressing AI training preferences. Metadata embeds rights-reservation tags directly into the files for images, videos, or PDFs. Natural language terms and conditions in a website can declare reservation of rights, but may not legally qualify as ‘machine-readable’.
The UK Government is currently evaluating how different copyright reform options would affect rights owners and GenAI developers. Four main options are being addressed: maintaining the status quo; requiring all authors to ‘opt-in’ by granting copyright licences for each work (supported by 88% of consultation respondents); a broad text and data mining (TDM) exception with no opt-out; or an opt-out model, whereby AI developers can use copyright-protected content unless the author actively reserves their rights.
No consensus has yet been reached by countries worldwide on copyright protection for wholly computer-generated work. Rather, there is a patchwork of legislative approaches to the issue, which means that universal enforcement of copyright in such works is problematic. In the UK, the issue is whether to amend special provisions in its CDPA, which currently give copyright protection to works ‘generated by computer in circumstances such that there is no human author of the work’. The author-cum-first copyright owner is designated as the undertaker of ‘the arrangements necessary for the creation of the work’ – this terminology is precisely the same as the CDPA uses to define a ‘producer’ in the context of determining an author/copyright owner of a film or a sound recording.
Copyright in computer-generated works endures for a fixed term of 50 years from the end of the year in which the work is made, as an exception to the normal length of the author’s life plus 70 years – because there is no human author. The current position of UK’s Intellectual Property Office (UKIPO) is that ‘it is unclear whether removing [protection for computer-generated works] would either promote or discourage innovation and the use of AI for the public good’; but does not rule out future changes.
In the US, where there is no copyright legislation expressly dealing with computer-generated works, the question is whether Congress should explicitly do so. The latest report by the US Copyright Office (USCO) addresses ‘Copyrightability’ of AI, and results from a two-year consultation with over 10,000 stakeholders, all 50 states and 67 other countries; and makes conclusions and recommendations. ‘Copyright protects the original expression in a work created by a human author, even if the work also includes AI-generated material. Copyright does not extend to purely AI-generated material, or material where there is insufficient human control over the expressive elements. Whether human contributions, to AI-generated outputs are sufficient to constitute authorship, must be analysed on a case-by-case basis. Based on the functioning of current generally available technology, prompts do not alone provide sufficient control. Human authors are entitled to copyright in their works of authorship that are perceptible in AI-generated outputs, as well as the creative selection, coordination, or arrangement of material in the outputs, or creative modifications of the outputs. The case has not been made for additional copyright or sui generis [stand-alone] protection for AI-generated content’.
USCO’s report extensively addresses the policy implications of providing copyright protection to AI-generated material, evaluates the arguments for and against legislative reform, and is likely to have worldwide influence. UKIPO will doubtless consider the report, and perhaps be persuaded to recommend amending the CDPA’s current protection for all computer-generated works that have ‘no human author’ – by removing protection only for wholly computer-generated works. In this way, authors using GenAI to produce only a starting point for new work, then adding their own creative expressions, would have copyright protection for those further original contributions to the final work.
During January and February 2026, independent research was commissioned by the UK’s Department for Culture, Media and Sport into ‘AI Copyright and Licensing in the Creative Industries’. The research examines the economic impact of different licensing and remuneration arrangements for creative content used by AI developers; and called upon views from artists, creators and freelancers. Research findings will inform a wider UK Government report on AI and copyright, scheduled for release by 18 March 2026, as required by the Data (Use and Access) Act 2025.
© Henry Lydiate 2026