Getty Images’ claim against Stability AI has the potential to be a watershed moment for the developing legislative framework – as well as public opinion – surrounding intellectual property (IP) and artificial intelligence (AI) systems in the UK.
Getty Images has issued a press release announcing that it has commenced legal proceedings in the High Court against Stability AI claiming “Stability AI unlawfully copied and processed millions of images protected by copyright and the associated metadata owned or represented by Getty Images absent a license [sic] to benefit Stability AI’s commercial interests and to the detriment of the content creators.”
Getty’s argument is that Stability AI has copied and processed copyright images for its AI art creator, Stable Diffusion, and that amounts to copyright infringement. Stable Diffusion is an AI model, trained on images found on the Internet, to generate artwork. The case will inevitably raise questions about the extent to which this is permitted under the fair dealing exceptions under English copyright law.
The UKIPO’s recent wave of consultations on the issue of IP and AI has brought to the foreground the tensions between copyright law and those exceptions, and hence between developers of AI systems (users) and those in IP-rich creative industries (rights holders). One outcome of those consultations was the UK government’s proposal to expand the current text and data mining (TDM) exception in copyright and database right law, beyond only non-commercial research purposes, to the mining of text and data for all purposes. The proposed expanded TDM exception would essentially enable users to trawl and mine otherwise protected works to “train” their AI systems without permission from, or payment to, rights holders.
There has already been public scrutiny of the government’s proposal to expand the TDM exception from the creative sector and, notably, the House of Lords Communications and Digital Committee . The Committee is of the view that the proposals go too far, and stated that they should be paused for further impact assessment, citing concern over the implications for businesses in the creative industries. “Developing AI is important, but it should not be pursued at all costs”, they say.
Thus, while there can be merit in AI systems having access to expansive datasets, it is clear that further thought is needed to ensure that the current proposal on TDM balances the positive aspects of AI with a clear steer on how rights holders will be properly protected from unfair exploitation, particularly in respect of those AI systems which churn out seemingly “creative” images, music and audio-visual content.
The dispute between Getty Images and Stability AI may give some insight into judicial thinking on the subject. It will no doubt also draw out considerable commentary from the government and a range of stakeholders, including the creative industries and the general public, as society continues to grapple with how it should balance the value of rapid technological innovation against the value of human creativity.