AI & copyright protection – What’s the latest?

Sarah Ezekiel suffers from motor neurone disease and sadly lost the ability to speak 25 years ago. Sarah uses Eye Gaze technology to communicate and speak with a synthetic voice, similar to that of the physicist Stephen Hawking. An old, scratchy VHS tape of Sarah playing with her children was found and an AI training model was used to replicate Sarah’s voice and accent. Sarah’s now grown-up children have now heard Sarah’s emotions through her own voice for the first time which is just incredible. Read the BBC article on Sarah’s story.

So where are we with AI from a legal perspective? Well, it was hoped that the UK’s Data (Use and Access) Bill (DUA) which passed in the House of Lords back in June would clarify the position. Disappointingly, we have to wait a bit longer to see how the Government is going to approach the regulation of AI and, in particular, data scraping and the training of generative AI.

DUA sets out the framework for use of new smart data regimes, as well as updating data protection laws. It has been a long time coming, with inputs and efforts from industry and numerous Government departments. The key issue is how to strike the balance between allowing AI systems to use data for innovation and training purposes, and protecting human creators’ works from being copied and used to train AI – at times without the creator’s permission.

DUA ping-ponged between the Commons and the Lords numerous times with both sides wanting considerable amendments. The Commons were in favour of innovation, arguing that an overly restrictive AI regime would scupper the UK economy. The House of Lords argued that organisations should be transparent about the information and proprietary works used to train AI models.  The latter view was supported by over 400 British musicians, writers and artists who signed a letter earlier this year, asking the Government to update copyright laws to protect original creators works from AI data scraping. Celebrities who supported the letter include Sir Elton John, Dua Lipa, Sir Ian McKellen, Sir Paul McCartney, Kate Bush, Coldplay and Richard Curtis.

The two houses could not come to an agreement and, so as not to delay the passing of DUA any further, the Government stated that the use of copyrighted materials to train AI would be considered separately under copyright consultations. For now, by March 2026, the Government will publish a full impact assessment report on the economic effects on the use of copyrighted works for AI training and data scraping, together with suggestions for possible statutory protections. A progress statement is expected by December 2025.

Although some may feel frustrated that we are no further forward as to the position of AI learning from creative works, the delay perhaps makes sense. The Government will undoubtedly want to consider the decision of the High Court in the eagerly awaited Getty Images v Stability AI case. The judgment, which is expected in the next few weeks, could significantly impact the intellectual property protections of original, human created works in respect of AI use and output.

Secondly, whilst this blog has focused on AI and copyright, there have of course been significant changes in respect of data protection laws including:

• The maximum fines for non-compliance, with rules around electronic communications and marketing, will significantly increase from £500k to £17.5 million or 4% of annual global turnover.
• The soft opt-in exemption will be extended to use by charities.
• Cookies notice/banners requirements are to be simplified, removing the requirement for user consent in specific circumstances.
• A new framework will be introduced which allows organisations to take a more pragmatic, risk-based approach to transfers of personal data outside the UK.
• Further safeguards will be introduced to ensure personal data is used legitimately including a list of examples of pre-approved legitimate interests which may be used as a lawful basis to process personal data.
• The Information Commissioners Office will become the Information Commission and the Information Commissioner will be known as the Chair.
• There will be specific legal mechanisms for sharing and reusing public sector datasets for research, innovation, and service delivery.

On the 27 December 2025, the EU’s Data Protection Board (EUDPB) is due to evaluate the UK’s adequacy decision after the EDPD granted a 6-month extension back in June 2025. Currently personal data can flow freely from the EU to the UK as the EU recognises that the UK’s data protection framework offers adequate protection for EU data subjects. The EUDPB granted this extension to give time, firstly, for the DUA to conclude its journey through Parliament and, secondly, for the European Commission to then evaluate the changes introduced by DUA. The introduction of any AI amendments (if they had ever been agreed!) would have resulted in DUA being delayed further and the UK having to request another extension from the EDPB (probably cap in hand).

We await the outcome of the Getty Images v Stability case with anticipation, as it will undoubtedly influence the Government’s approach to the use of AI in respect of copyrighted works. It will be interesting to see whether the Government and its suggested copyright updates will take a liberal view to AI data use and training, an approach adopted by the USA for innovation and economic reasons. The Government, of course, may introduce transparency obligations in order to balance innovation and the interests of proprietary rights- the stance taken by the EU under its AI Act. Perhaps a middle ground could be a copyright licensing regime whereby AI developers are able to use human developed works in return for a licence fee.

As to the legal position of training AI on human created works, we shall have to wait a little longer. Whatever the outcome, AI training doesn’t have to be seen as a bad thing. If conducted the right way, AI can be used and trained for meaningful purposes, look how it changed Sarah’s life.

For support with use of AI, intellectual property rights and data protection, please contact Abi Sinden on [email protected] or call 01202 306273.