Data (Use and Access) Bill [HL]

Baroness Stowell of Beeston Excerpts

Report stage
Tuesday 28th January 2025 Lords Chamber

These amendments are needed to ensure that the AI companies adhere to the copyright law and, in the process, ensure the future of our world-beating creative industries. If the noble Baroness, Lady Kidron, calls a vote, I will be voting for Amendment 44A, and I encourage other noble Lords to join me in the “Contents” Lobby.

[Baroness Stowell of Beeston (Con)]

My Lords, this topic understandably arouses a lot of emotion, but it is a difficult one to resolve satisfactorily. The Communications and Digital Committee has examined the challenge of copyright in an AI world from several angles over the last couple of years, and our conclusion as a result of that work is that the tech and creative industries need to find a mutually beneficial way forward on copyright because, in this new world, they are relying on each other to succeed. The AI models and services that the tech platforms are building for consumer and commercial use, such as Chat GPT, Claude, Llama, Grok and others yet to emerge, have an insatiable and ongoing appetite for new, quality data and original content, and it is a continual supply of that content which will make them yet more sophisticated, and how each platform operator will compete in the race to dominate. What I have just described is also why the Government should not pursue copyright laws that primarily benefit foreign tech firms that are prepared to pay vast sums for energy, computing facilities and staff, but not, as we have heard, for data.

During our inquiry on large language models, we heard contrasting interpretations of existing copyright law. Our view is that the application of copyright law in the context of AI is complex, but the principles remain clear. What is needed is a framework that aligns incentives between content creators and AI firms to help them strike mutually beneficial deals. In our reports, we have called for that framework to include: a transparency mechanism to allow rights holders to check for infringements; much better technical and legal enforceability; and measures to support a new market in responsible AI training data.

I am pleased to say that the amendments in this group from the noble Baroness, Lady Kidron, and my noble friend Lord Camrose, reflect these objectives, but I would like to make some further comments on the way forward. While the Communications and Digital Committee has welcomed the Government’s copyright consultation as a step forward in making progress on this issue, we have cautioned strongly against adopting a flawed opt-out regime comparable to the version operating in the EU. Indeed, Matt Clifford’s recommendation that we adopt that EU model is the only part of his excellent AI Opportunities Action Plan that I disagree with.

That said, ensuring the UK remains competitive in this global market is vital, and some might argue, contrary to what the noble Baroness has said, that the arrival of DeepSeek brings that into sharper focus. It is why I suspect the Government prefer an opt-out model. What we as a committee argue is that if, after their consultation, the Government decide to go ahead with an opt-out model, it must include the transparency, technical and stronger enforcement mechanisms I have already outlined and that are reflected in these amendments. What is important therefore to understand is that the amendments in this group could apply to an opt-in or opt-out model; they are flexible.

Whichever route the Government take, it is essential that, alongside the creative industries that we have heard are so important to our economy and society, the conditions are set for our domestic AI tech sector to scale and compete. UK spin-out and start-up innovators can seriously challenge existing dominant tech firms with specialist AI models and new services and applications. Not only must our copyright regime not be a barrier to entry for UK start-ups seeking to scale but the UK needs a workable framework to incentivise a dynamic licensing market to promote and seize the economic value of the high-quality data this nation holds. That could make this country an attractive AI training destination for all AI models.

Sorting all this out is urgent, and it is not easy. I worry that if we do not resolve it soon, the UK will be defined by our concerns about copyright to our detriment in the AI global race. The Government cannot wait for the courts to find a way forward; they must act swiftly once the consultation is over. This Bill is the right vehicle for doing so, and because it is what we call a Lords starter, it is yet to go through all the Commons stages; we are at the start of this process, and the Government have time.

For all the reasons I have outlined, if the noble Baroness divides the House, I will support her in voting for these amendments.

[Lord Freyberg (CB)]

My Lords, as one of the supporters of these amendments, I support the amendment so expertly moved by the noble Baroness, Lady Kidron. I declare my interest as someone with a long-standing background in the visual arts and as an artist member of DACS, the Design and Artists Copyright Society.

I thought it would be helpful to highlight and focus on just one element of the noble Baroness’s speech, specifically the issue of transparency. Here, there is a theme developing throughout the House on this issue. One of the biggest obstacles to ensuring fair pay for creators is that AI companies have not been transparent about what works. They have been used for training AI models. Tech companies have rebuffed transparency measures because they say that this will reveal trade secrets. While I understand that business need, it cannot come at the expense of creators. There is a way in which to make transparency measures work for both business and creators, giving access to creator representatives about the use of their work on a confidential basis to facilitate copyright licensing.

This is, after all, what data rights have done for millions of people, giving them the agency to know when their data has been used. It is entirely reasonable and possible for transparency measures to be upheld and properly enforced. Therefore, considering the significance of this issue, I should be very grateful if the Minister will confirm that transparency measures proposed in the copyright and AI consultation will not be conditional on a reservation rights system.