(2 weeks ago)
Lords ChamberMy Lords, I pay tribute to the Lords committee that has considered this issue. We are keen to make progress in this area but it is important that we get it right. The previous Government had this on their table for a long time and were not able to resolve it. The Intellectual Property Office, DSIT and DCMS are working together to try to find a way forward that will provide a solution for creative media and the AI sectors. Ministers—my colleagues Chris Bryant and Feryal Clark—held round tables with representatives of the creative industries and the AI sector recently, and we are looking at how we can take this forward to resolve the many issues and questions that the noble Lord has quite rightly posed for me today.
My Lords, in the absence of legislation or policy it is left to the courts to wrestle with whether the training of large language models infringes copyright. For over two years now, cases have been progressing in the US, in Europe and in the UK. A German first-instance decision recently found in favour of the machines—it will be appealed. Noting my interests as an IP litigator, do the Government believe, as I do, that English common law is capable of resolving this tricky issue?
My Lords, the Government are clear that copyright law must be respected when content is used to train AI models. If copies are made of protected work, licences must be required from the copyright owner unless a specific copyright exception applies. The problem is that the law does not yet apply equally to generative AI models, and that is the issue we are grappling with. Our view is that this should not necessarily be left to the law; unfortunately, it takes a long time for these legal cases to be resolved. We are trying to find a way forward that will be fair to everybody but that does not require the long legislative process that I know the noble Earl is all too aware of.