Get in touch
If you have any questions relating to this article or have any intellectual property matters you would like to discuss, please contact us.
Update, 25th June 2025: In its closing submissions in the High Court, Getty withdrew its main copyright infringement claims, due to evidential challenges. In light of this, responsibility for clarifying how copyright law and AI innovation will co-exist rests firmly with the Government, which is expected to come under pressure to respond sooner than the timetable agreed in passing the Data Use and Access Act 2025.
Last week heralded two significant moments for the legal landscape of AI and intellectual property. The Data Use and Access Bill (DAU Bill) was finally passed, ending its extended passage through Parliament. Meanwhile in the High Court, the case of Getty Images v Stability AI began a long-awaited 18-day hearing.
Both developments have set the spotlight on the use of copyrighted material for training AI models, and consequential issues of rights ownership and potential for infringement in AI-generated works. For creatives and rights owners, these are important and emotive topics; Tom Stoppard, Paul McCartney, Dua Lipa and Elton John were among the high-profile artists to speak up on the DAU Bill. Sir Elton called it an “existential issue” for artists and urged the Government “to do the right thing.”
The Copyright, Designs, and Patents Act 1988 confers copyright protection on original works and the copyright owner has exclusive right to reproduce, adapt and distribute those works. The ‘intelligence’ that powers AI models relies on vast data sets, which are derived from pre-existing content and materials. This input data may be public, third party owned, or user generated. Where that data benefits from copyright, its use for training an AI model is likely to infringe the relevant owner’s rights. AI outputs are inherently derivative and, if an AI-generated work reproduces or adapts a “substantial part” of a protected work (without a licence) further infringement can arise.
Both providers and users of AI models may be liable for infringement, but for rights holders the challenge is transparency – without evidence of how a model has been trained, proving infringement may be difficult. This, and creatives’ desire for fair compensation for their artistic effort, lies at the heart of the tension between AI innovation and creative rights.
In recognition of this tension, the Government’s consultation on Copyright and Artificial Intelligence set out four possible options to govern the use of copyrighted materials in training AI: leave the current laws unchanged; require developers to obtain appropriate copyright licences; permit use unless rights holders reserve their rights; or create a new exemption under copyright law for AI training. The Consultation received over 11,500 individual contributions; a Government response is awaited.
Meanwhile, the AI Opportunities Action Plan makes clear the Government’s ambition to “shape the AI revolution on principles of shared economic prosperity, improved public services and increased personal opportunities”. The report emphasises the need for AI-developers to have access to ‘high-quality data’ to support innovation.
It’s against this backdrop that Baroness Kidron introduced Amendment 49 to the DAU Bill, proposing transparency requirements to oblige AI developers to disclose the data used in training their models. In Europe, transparency requirements already exist under the EU’s AI Act. This will require developers to publish summaries of copyrighted data used for training and to comply with EU copyright law.
The Government argued it was not appropriate to deal with this issue in the DAU Bill, the primary purpose of which is data privacy reform. After a protracted ‘ping-pong’ between the Houses, the Lords ultimately approved the DAU Bill without including the transparency requirement. The concession was made following an amendment proposed by the Commons, which mandates that the Secretary of State for Science, Innovation and Technology must publish a full technical report on its copyright and AI proposals within nine months of the DAU Bill receiving Royal Assent (and an interim report within 6 months).
The outcome of Getty Images v Stability AI will have significant ramifications for the issue and may significantly inform future legislative direction. Getty claims the training, deployment and outputs of Stability AI’s Stable Diffusion model give rise to trade mark infringement, passing off, copyright infringement, and database right infringement. A parallel case is under way in the US courts. Getty’s legal team have emphasised that their case is not seeking an end to AI. Rather, they believe creatives and technology “can exist in synergistic harmony” but believe “the problem is when AI companies such as Stability want to use those works without payment".
For now, we wait for the Government’s report and response to the Consultation, the High Court’s ruling in Getty. In the meantime, businesses have plenty of data considerations to engage with from the DAU Bill. The majority of the reforms rely on secondary legislation, however amendments to UK GDPR and Privacy & Electronic Communications Regulations are expected to come in quickly.
If you need advice on your intellectual property rights, use or AI or how the DAU Bill will impact your business, please get in touch with our Commercial and Technology lawyers at [email protected].
Share:
If you have any questions relating to this article or have any intellectual property matters you would like to discuss, please contact us.
Subscribe to our newsletters where our experts share insights on topics tailored to you.
Sign up to our newsletter