Dr Allison Gardner

Perfect. I asked the Secretary of State what reassurances can be given that smaller creatives, including microbusinesses and small creative businesses, will be considered in the report so that they can have confidence that the systems finally applied will work for them, particularly when we consider an individual’s early career—think of Ed Sheeran strumming away in his bedroom in his pre-fame days—and how they can protect their copyrighted works against the global tech giants. New clause 16 addresses the economic impact on both copyright owners and AI developers, and I want to switch from talking about copyright owners to trying to defend the AI industry. If we do not get the controls right, we risk the mid and long-term success of the AI industry. If we do not get a fair solution for the creative and AI industries, we risk a reduction in the quantity, and potentially in the quality, of human-created data and an increase in AI-generated creative data. I will briefly segue, because we are developing a lot of AI-created content that might be subject to copyright. A report recently pointed out that 18% of Spotify content is now AI-generated. People might remember the big hoo-ha when an AI-generated image won a photographic competition, which caused a lot of disturbance, but a lot of creative skill was involved in how the photographer developed and produced that image. No, it was not a photograph, but it is in a category of its own. I feel that is also creative content and copyrighted data, so there is a grey area. If we start to generate more and more AI-created data and less and less high-quality human-generated data, because of the challenges to the creative industry, there is a danger that AI models will start scraping and training on AI-generated data, potentially leading to a reductive spiral into mediocrity, with some even suggesting that this could result in model collapse. On new clauses 16 and 17, I encourage the House to consider the impact of not employing proposals such as licensing and protecting the generation of new human-created content, given the risks posed to AI models and developers in the long term. I will briefly comment on amendments 37 and 38, tabled by my hon. Friend the Member for Newcastle upon Tyne Central and West (Chi Onwurah). She ably outlined the reasons for the amendments, so I will not go into a lot of detail, but I want to point out that getting the definitions correct will prevent a loophole whereby AI companies can misuse personal data by claiming that their commercial development is scientific research. The amendments would provide transparency on the use of data by researchers in order to maintain confidence in this country’s ethical, legal and professional high standards in academic research. I hope the Minister will give careful consideration to the points I have raised. I am now going to give my Whip, my hon. Friend the Member for Cardiff North (Anna McMorrin), a heart attack because I am going to refer to amendments 41 to 46 to clause 80 on article 22 of the UK GDPR, which is close to my heart. She is not to worry, though; I read those amendments with great interest and I understand the back-up they would provide, but although I am a newbie MP, as I read them—in my understanding, given the little work I did in my previous job with a regulator—I felt that they were more like secondary legislation. They could be considered for the future, particularly amendment 46, which includes some very welcome additions. However, when it comes to primary legislation, I feel that the Bill works better as it stands.