Digital and Technology Policy: National Sovereignty

Lord Holmes of Richmond Excerpts

Thursday 23rd July 2026 Lords Chamber

[Lord Holmes of Richmond (Con)]

My Lords, it is a pleasure to take part in this debate. In doing so, I declare my relevant technology interests as adviser to the Crown Estate and to Simmons & Simmons LLP.

I must start by congratulating my friend, the noble Baroness, Lady Kidron, on her brilliant and blistering introduction to this debate. The noble Baroness is having quite a month of it: this debate today, a new book out and a fantastic White Paper published just last week. I know it is only a matter of time—and quite right—before moviegoers will be treated to “Baroness Beeban: The Biopic”. I will be speaking with Christopher Nolan later today.

The departmental chairs have been rearranged. What does this mean, though, for digital sovereignty? Yesterday, the Minister, Jonathan Reynolds, described the new department as a “genuinely new beast”. Make of that what you will, but largely, the technology industry was dismayed that DSIT is no more. I ask the Minister: can they all be wrong?

We also have GDS on the move. Rather than being the positive, impactful, cross-departmental force, it will now be within DCMS, so it will be part of a department, giving a departmental view with a departmental Secretary of State, seeking to influence other different government departments. How will this play out?

Similarly, we have the separation of AI sovereignty and digital infrastructure, now relying on co-ordination where before we had effective oversight. How will this benefit digital sovereignty?

We have the AI Security Institute now responsible within that office for increasing take-up of technology across government. If you have the same group responsible for testing the tech and deploying and supplying the tech, ask anybody what the outcome will be and you will get a single—and in this context, unhelpful—answer. Add to that what is happening with AI IP and copyright and AI growth zones and you have quite a cocktail, none of which on the face of it seems to have an effective and laser focus on digital and AI sovereignty.

So, what do we require? As is always the case, I take your Lordships back to the three Cs: clarity, consistency and coherence of approach. It does not matter whether you are a citizen, creative, consumer, innovator or investor: it is about clarity, consistency and coherence. In the specific context of digital sovereignty, it is about definitions, dependencies and decisions. There are as many definitions of sovereignty as you care to take, but the Government need to pick one, stick to it and clearly and coherently communicate it.

On dependencies, what are the Government prepared to tolerate? We do not need to—nor could we—do full-stack sovereignty. But the truth of it is that no nation, not even the United States of America, nor China, has true, pure, full-stack sovereignty. It is not doable and it is not necessary. What is the risk profile that the Government are prepared to tolerate, and what do they say in terms of broad resilience? To govern is to choose: ultimately, these decisions have to be made.

We have heard variously from the Minister over debates and Oral Questions in your Lordships’ House that digital and AI are now across all domains. Nobody would disagree with that, but would that not suggest that we need a cross-departmental, cross-domain approach when it comes to our legislation and regulation? That is the approach being taken with the cyber Bill, and I agree with it—although it is deficient and absent in some serious areas, with no mention currently of AI, and we will be coming back to that in the September sitting. If it makes sense to have a cross-sector cyber Bill, surely it makes sense to have a cross-sector, cross-economy AI Bill.

For digital sovereignty and AI sovereignty we must think cross-sector and be principles-based and outcomes-focused, with inputs understood. Those principles include trust and transparency, inclusion and innovation, interoperability and an international outlook, accountability, assurance and accessibility. Many of those principles have been weakened by the recent changes to departmental responsibilities, and there needs to be that laser focus on clarity, consistency and coherence. Ultimately, digital sovereignty, AI sovereignty and data sovereignty are all essential elements, and we need to look at how we take an approach which enables, empowers and fundamentally allows all of us as citizens to say, with full throat, that our data, our decisions and our digital futures, together, are human-led, inclusive digital futures.

Artificial Intelligence: Legislation

Lord Holmes of Richmond Excerpts

Thursday 16th July 2026 Lords Chamber

[Baroness Lloyd of Effra (Lab)]

My noble friend is right that AI is transforming the military and defence. It is indeed incredibly important that we work internationally. Last week, the UKAC co-ordinated the most recent meeting of the International Network for Advanced AI Measurement, Evaluation and Science, where we are working together with other countries’ ACs on best practices for model evaluations, using that expertise to build tangible international consensus to mitigate AI risks.

[Lord Holmes of Richmond (Con)]

My Lords, I declare my technology interests as set out in the register as an adviser to the Crown Estate and to Simmons & Simmons LLP. I thank the Minister for her response to my Written Question of 8 July, setting out what the Government are doing on the international stage, not least with the Council of Europe AI framework. But does the Minister agree that if the UK Government take a point-of-use, sector-by-sector approach then there will be accountability deficits, transparency difficulties and a lack of the three Cs that are crucial whether you are an investor, an innovator, a creative or, crucially, a citizen when it comes to AI: clarity, consistency and a coherent approach? Would an AI Bill not resolve that for the benefit of us all?

[Baroness Lloyd of Effra (Lab)]

It remains the case that we must be very attuned to developments in AI, in AI risks and in AI opportunities. We are extremely fortunate to have the AI Security Institute, which can provide us with that advice. Obviously, we will always look at whether there are regulatory or legislative gaps and we will act if we need to.

Cyber Security and Resilience (Network and Information Systems) Bill

Lord Holmes of Richmond Excerpts

2nd reading
Tuesday 14th July 2026 Lords Chamber

[Lord Holmes of Richmond (Con)]

My Lords, it is a pleasure to take part in this Second Reading debate and to follow my friend, the noble Viscount, Lord Colville of Culross. Though he, by his own words, repeated some of the earlier points, he was the first speaker to say “lacunae”, for which I am particularly grateful; it sounds like a technology company. I declare my technology interests as set out in the register, as advisor variously to the Crown Estate and to Simmons & Simmons LLP.

As has been noted, this Bill is significant by having “cyber” in the title. This is long overdue, much needed and critical, as the Minister said, taking a cross-sector approach to cyber. Yet the first interesting point is that the Bill does not do that. Most notably, it is extraordinary that neither food nor space are included in the Bill. Similarly, it is said to take a cross-economy approach, yet it does not. However, it is worth mentioning the two sides of that economic coin and the huge economic growth potential from our cyber industries. I echo all the points that have been made about the need for skills and education, and to enable the cyber sector to grow and deliver that economic benefit.

Reporting has been mentioned widely throughout the debate. It is unfortunate if one finds themselves in a situation such as that of my noble friend Lady Harding, with multiple agencies to have to report to. Surely it would make sense to have a single reporting point for the speed, efficiency and effectiveness of that reporting system. Similarly, there is so much opacity around many of the definitions within the Bill. I pull out “significant impact” as one of them. What does this mean? If it stays as set out, the natural and understandable response from business is to go for the low-level mass reporting to avoid regulatory intervention. Does that enable the economy and the country to be better protected in this respect? I think not.

Similarly, the 24-hour and 72-hour reporting requirements feel oddly constructed around the artificial concept of a day set out in 24 hours. When one considers the real-time velocity of these attacks, it would seem logical that something way ahead of 24 hours would be advantageous at that stage and potentially something extending beyond 72 hours to do the second bit of the reporting process. What evidence is there to support this quite arbitrary 24 hours—or, as otherwise described, a day?

Security and resilience are the concepts most central to the Bill and will be the markers of its success or otherwise. For issues around proportionality and the ability to evolve and develop, does the current structure of the Bill really optimise this? MSPs have already rightly been mentioned. The burden for them is overbearing as currently set out. Surely it would make more sense to have a concept around what is reasonable for MSPs to oversee, what can legitimately be seen as within their control and how they can evidence that.

Understandably, the multiple regulator issue has been well discussed. It is critical, because how will we have co-ordination across all those organisations? A forum is certainly not the solution, as my noble friend Lady Harding rightly set out. A lead regulator, a single regulator or something around that has to be the way to go. I would argue that the NCSC should have the loudest voice in determining what is the best model and the best structure for doing that.

I was interested in the Minister’s introduction. She talked about the criticality of cross-sector consistency. I agree entirely, but in a debate on AI on 4 June in Grand Committee, she argued that consistency was not necessarily a central principle for the Government in the regulation of AI. I believe that a true cross-sector approach to cyber makes sense in this Bill. We can add to what is currently there and make that happen, but it is surely logical, and indeed a consistent approach to consistency, that we take that approach with AI. Can the Minister say why, if this argument is good for cyber, it is not good for AI—with a cross-sector approach, as will benefit this Bill, that is principles based and outcomes focused, with inputs understood?

I have another point on consistency. When the Minister sums up, can she set out the advantages that she sees in all the divergences that the Government have taken from the EU’s NIS2? I would be interested to hear the Government’s arguments for the advantages that they are seeking to bring from that.

My noble friend Lord Arbuthnot has rightly mentioned the CMA, and I know that my friend, the noble Lord, Lord Clement-Jones, will also mention this. It is right to mention it in this Bill, even though the national security Bill is said to be the vehicle through which this will come forward. It is right to give it a run around the track in the legislative process with this Bill, because we are talking about coverage currently holding our cyber professionals back because of a 1990 statute. To give some sense of what that means, in 1990 it had been only 24 years since England had won the World Cup. It is in urgent need of reform. We need to empower and enable our great cyber security professionals to do their job.

I have a quick point on DVS. I am not sure the Minister was involved when we did the Data (Use and Access) Act, but there was rightly a lot of discussion around DVS. In some ways, parts of this Bill are the other side of that coin. Currently we have a situation that is not addressed in the Data (Use and Access) Act or in this Bill: what happens to verification services when they have an attack where synthetic data is injected directly into the data stream, in effect bypassing the camera to get verification? How does this Bill address that issue? Do Clause 12, the potential regulations under Clause 30 or the guidance under Clause 36 address this? If not, what is the Minister’s view as to how we address that critical issue around synthetics?

The success of the Bill and of cyber in the UK will rest so heavily on the shoulders of our cyber security professionals, the women and men who do so much to keep our system—and, through that, us—safe. We owe them so much. Understandably, they often do this in the shadows, in dark rooms and in the Doughnut. We give them our sincere thanks, and we must demonstrate that thanks through the amendments we bring forward to make the Bill better.

Open Artificial Intelligence Service

Lord Holmes of Richmond Excerpts

Wednesday 3rd September 2025 Lords Chamber

[Lord Vallance of Balham (Lab)]

A lot is going on. What is being considered is investment in companies, but things that have happened include the formation of a new AI computer infrastructure—Isambard-AI went live this summer. The AI Security Institute is determined to make sure that we have the ability to look at risks of future models, and of course there is the National Data Library. So, there is a lot in train and a lot of sovereign capability is being built. The next phase is to make sure that we have investment for companies that are developing the latest models in the UK.

[Lord Holmes of Richmond (Con)]

My Lords, I declare my interests as set out in register, not least as adviser on AI to the Crown Estate and Endava plc. Public engagement and public trust are critical to the success of sovereign AI—indeed, critical to the success of all AI. What are the Government doing to engage the public with their sovereign AI aspirations, and to build on that public engagement? Can the Minister inform the House when the consultation paper on the proposed AI Bill will be published—September?

[Lord Vallance of Balham (Lab)]

The sovereign AI unit is already publishing what it does on its website and consulting very broadly. As the noble Lord knows, there is a consultation on the AI Bill. I can confirm that it will not be September, but I cannot confirm beyond that.

Artificial Intelligence: Legislation

Lord Holmes of Richmond Excerpts

Monday 21st July 2025 Lords Chamber

Asked by

[Lord Holmes of Richmond]

To ask His Majesty’s Government when they will publish their consultation on their plans for artificial intelligence legislation and when they expect any subsequent bill to be introduced.

[Lord Holmes of Richmond (Con)]

My Lords, I beg leave to ask the Question standing in my name on the Order Paper and declare my technology interests, as set out in the register. They include being a member of the global advisory board at Endava plc and a member of the science and technology advisory committee of the Crown Estate.

[The Minister of State, Department for Science, Innovation and Technology (Lord Vallance of Balham) (Lab)]

My Lords, the Government are preparing a consultation on AI legislation in order to gather views on the proposals. This would better prepare the UK for AI security risks, while making sure that our statute book is ready for the age of AI and its undoubted opportunities. The Government will update Parliament in due course.

[Lord Holmes of Richmond (Con)]

My Lords, I was rather hoping that we might have a consultation for our summer reading, but we will await the consultation and subsequent Bill. The Government have said that they will take a domain-specific approach to the legislation and regulation of AI, rather than cross-sector. To that end, how will consistency be assured through such an approach?

Similarly, what about areas that currently do not have any competent regulator, such as hiring and recruitment? People find themselves not being shortlisted for roles because AI has made that decision, without even knowing that AI was in the mix. Even if they knew that AI was in the mix, there would be no place to seek redress. Surely, clarity, certainty and consistency are what anyone in the country requires when it comes to AI, whether they are an investor, innovator, consumer or creative. How will a domain-specific approach, with no guiding mind, ensure that clarity, consistency and certainty?

[Lord Vallance of Balham (Lab)]

I am sorry that I disappoint the noble Lord with his summer reading list. I am happy to meet him to give him some other recommendations of good books.

As set out in the AI Opportunities Action Plan, we believe that most AI systems should be regulated by the existing regulators. They are the experts. They need the AI skills to be able to do it. The Government are working with regulators to drive collaboration and alignment across the regulatory domains through, for example, the Digital Regulation Cooperation Forum’s AI and digital advisory hub and the Regulatory Innovation Office, which is working with DRCF to collaborate on the support of the development of tools to help businesses and investors better navigate digital regulations.

We recognise the point the noble Lord has raised, which is that there are some aspects of AI that need to be looked at across AI generally. That is why we are undertaking consultation on legislation, and why we have ongoing work with all the departments around the impact on jobs that he described.

AI Opportunities Action Plan

Lord Holmes of Richmond Excerpts

Thursday 5th June 2025 Lords Chamber

Asked by

[Lord Holmes of Richmond]

To ask His Majesty’s Government what steps they have taken to implement the recommendations of the independent report AI Opportunities Action Plan, published on 13 January.

[Lord Holmes of Richmond (Con)]

My Lords, I beg leave to ask the Question standing in my name on the Order Paper and declare my technology interests as set out in the register.

[The Minister of State, Department for Science, Innovation and Technology (Lord Vallance of Balham) (Lab)]

The Government are taking action to capitalise on AI’s potential and welcomed the publication of the AI Opportunities Action Plan, accepting all 50 recommendations. Implementation is well under way. We have launched the AI growth zone application process, held the first meeting of the AI Energy Council and signed an MoU with Anthropic. We are delivering the AI research resource, including the Isambard-AI and Dawn supercomputers, which will boost the UK’s AI compute capacity thirtyfold.

[Lord Holmes of Richmond (Con)]

My Lords, Matt Clifford delivered an excellent report, with 50 wide-ranging recommendations across our economy and society. Does the Minister agree that the fact that they rightly range widely makes clear the need for the Government to bring forward cross-sector AI regulation to ensure that, wherever we come across AI in our lives, there will be clarity, certainty and consistency on how we have that AI experience, which would surely be good for innovators, investors, creatives, citizens and our country?

[Lord Vallance of Balham (Lab)]

There are three approaches to making sure that we get consistency and appropriate regulation and support, as the noble Lord suggested. The first is that the regulators look after AI in the domains which they already look after. We are making sure that they are properly supported to do that and can join up—for example, in the Digital Regulation Cooperation Forum—to learn lessons across different areas as they apply AI in their domains. The second is the approach of assurance: to develop an assurance industry in the UK which can assure people that, when they use AI, it performs what they expect it to and in the way they expect. That is true both for the Government—the Artificial Intelligence Playbook for the UK Government addresses some of the wider issues—and in developing the assurance industry overall and looking at areas such as bias in systems. The third, as he alluded to, is the question of what happens as artificial general intelligence, artificial superintelligence and the latest models come along. We remain committed to bringing forward AI legislation so that we can realise the enormous benefits and opportunities of this technology in a safe and secure way. We continue to refine our proposals and hope to launch a public consultation before the end of the year.

AI: Cross-sector Legislation

Lord Holmes of Richmond Excerpts

Tuesday 29th April 2025 Lords Chamber

Asked by

[Lord Holmes of Richmond]

To ask His Majesty’s Government whether they plan to introduce cross-sector legislation on artificial intelligence in 2025.

[The Minister of State, Department for Science, Innovation and Technology (Lord Vallance of Balham) (Lab)]

The Government remain committed to bringing forward AI legislation to realise the enormous benefits of this technology in a safe manner. We are continuing to refine our proposals to deliver this, ensuring that they are proportionate and incentivise investment and innovation. We will launch a consultation later this year. Most AI systems are already regulated at the point of use by the UK’s existing regulators. In response to the AI action plan, the Government are committed to working with regulators to support them in boosting their capabilities.

[Lord Holmes of Richmond (Con)]

My Lords, last month I published a report making the case for cross-sector AI legislation. Is it not clear that AI is already impacting, positively and otherwise, cross-sector, cross-society and cross-economy? If we do not have a cross-sector approach through legislation, how will we enable the clarity, the certainty and the consistency of approach which will bring forward the confidence to enable innovation and investment—good for citizen, good for consumer, good for creative, good for British business and good for our country?

[Lord Vallance of Balham (Lab)]

I thank the noble Lord for his question and enjoyed reading his report very much. There are three ways in which this cross-government AI approach will be looked at. First, as I say, the existing regulators will regulate their own areas. They will also be brought together more. The digital forum already brings together regulators around AI and has been given more money to ensure that the regulators can join up on this. Secondly, the development of assurance tools, which has been outlined in the AI Opportunities Action Plan, will allow us to understand that the actual use of AI is using tools that are validated. There will be a market in making sure that the validation system grows and becomes an important way of assuring users. Thirdly, the consultation around the newer models —advanced general intelligence and superintelligence, as it arrives—will require a cross-cutting general piece of work, which is where the consultation starts later this year.

Artificial Intelligence: Emissions

Lord Holmes of Richmond Excerpts

Wednesday 23rd April 2025 Lords Chamber

[Lord Vallance of Balham (Lab)]

Not being from DESNZ, I will ask somebody from that department to give the noble Earl a specific answer on that project, but we will look for where there are sources of energy as we think about where to place data centres. The recent call for expressions of interest in AI growth zones has had more than 200 expressions of interest. They will be considered on the basis of where the energy is available, or could be available, and how we can ensure that we get a clean energy supply to data centres.

[Lord Holmes of Richmond (Con)]

My Lords, will the Minister consider leading a review into the PUE measure for data centres? The power usage effectiveness measure has been around for some time now. Does he consider it to be effective and the optimal way of measuring the impact of data centres from an energy perspective?

[Lord Vallance of Balham (Lab)]

I thank the noble Lord for that important question, on which I have had discussions with him before. I think it is an important way of measuring it, but the new AI Energy Council is looking at all forms of link between energy and AI. It is worth also noting that the advances in technology mean that the energy consumption is dropping for many of the approaches to new compute and data centres. In fact, some of the chips being designed now may reduce the energy consumption between hundredfold and thousandfold. There is a need to keep an eye on this and to think about what the appropriate way is to measure both the energy consumption and, as the noble Lord rightly says, the broader environmental impact, including water usage.

Artificial Intelligence: Regulation

Lord Holmes of Richmond Excerpts

Monday 10th February 2025 Lords Chamber

Asked by

[Lord Holmes of Richmond]

To ask His Majesty’s Government, following the publication of the AI Opportunities Action Plan on 13 January, what plans they have to introduce legislation in 2025 for regulating artificial intelligence in areas including intellectual property, automated decision-making, and data labelling.

[Lord Holmes of Richmond (Con)]

My Lords, I beg leave to ask the Question standing in my name on the Order Paper and declare my technology interest as set out in the register.

[The Minister of State, Department for Science, Innovation and Technology (Lord Vallance of Balham) (Lab)]

As set out in the manifesto, the Government are developing legislative proposals which will allow us to safely realise the benefits of the most powerful AI systems. The Government are also consulting on AI and copyright. Next steps, including on potential legislation, will be decided once all evidence is considered. The Data (Use and Access) Bill will enable the responsible deployment of solely automated decision-making, with stringent safeguards in place for individuals.

[Lord Holmes of Richmond (Con)]

My Lords, with the world talking AI in Paris this week and with parts of the EU AI Act already brought into force earlier this month, is it not over time for the Government to bring forward AI legislation in the UK: for the jobseeker who constantly finds herself not making the shortlist, not even knowing that AI is in the mix, or for the creative constantly finding her work stolen by AI with no consent, no remuneration and no respect? Does the Minister agree that sector-wide AI legislation, ushering in right-sized regulation, is good for investment, good for innovation, good for creatives, good for citizens and good for all our AI futures?

[Lord Vallance of Balham (Lab)]

As the noble Lord points out, getting regulation right here is good for investment and good for business. We are taking the approach of regulation by the existing regulators for the use of AI. We intend to bring forward legislation which allows us to safely realise the enormous benefits of AI in the frontier space. Of course, in the Data (Use and Access) Bill, some of the issues the noble Lord raised are already addressed.

Creative Industries: Rights Reservation Model

Lord Holmes of Richmond Excerpts

Thursday 30th January 2025 Grand Committee

[Lord Holmes of Richmond (Con)]

My Lords, it is a pleasure to follow my noble friend Lord Black. I congratulate the noble Lord, Lord Foster, on securing this timely and excellent debate. In doing so, I declare my interests as set out in the register—in particular my technology interests, not least as an adviser to Socially Recruited, which is an AI business.

As the noble Lord, Lord Foster, has already set out, we had an excellent debate on Tuesday night. My question for this afternoon is: how much does it cost to develop and train a foundation model? Is it £500 billion or £5 million? Is it somewhere in between? I do not know, but here is what we do know. The cost of current foundational models is felt by our creatives: the musicians who make sounds where there would otherwise be silence; and the writers who fill a blank page with words that touch our human hearts and souls and, sometimes, change the course of human history. They are paying the cost of the current “model” that we have.

How can it be not only that they are currently footing the cost but that the potential, proposed approach to this issue will put the onus on them to assert their rights? There is that onus, the cost, pressure and stress and, ultimately, the impossibility of doing this with an opt-out model. My first question to the Minister is: can it ever be so that opting out could work? How could it ever bring the certainty, clarity and consistency that we require? As a helpful example, can the Minister say something about the recent LAION case and the light that that throws on this matter?

There is a real tedium to this TDM discussion. It is just that an obvious and irrefutable truth is wilfully ignored and pushed to one side. If you own a copyright or have IP rights, you hold and own those rights. If you do not, the truth is simple and unquestionable: those rights are not yours. That should be the guiding principle when considering any potential approach to IP and copyright in relation not just to AI but to the fact that we have hundreds of years of legal certainty which comes from this.

How would the Minister define a proper and workable model for the preservation of these rights? What would he say to individuals and small entities about the cost, pressure and impossibility of seeking to enforce their rights? How does he intend transparency to be an important thread that runs through this alongside the technical? What about post-ingestion and, if we get to the point of some potential change, what about all that protected material already ingested deep into the engine room of these models?

What attracts businesses, investors and innovators to the UK from a regulatory and legislative perspective? It is certainty, clarity and consistency. In no sense can we say that we have those right now in our country. That is why I believe, not only when it comes to IP and copyright, that given all the issues we are currently grappling with in these new technologies, not least AI, we should have overarching AI legislation and right-sized regulation, which is always good for all elements of our economy and society. Yes, look at IP and copyright, but we should have an AI authority with AI-responsible officers labelling sandboxes and, crucially, a complete transformation of public engagement.

It seems clear at this stage that when it comes to the Government’s plans for IP and copyright in relation to AI, we should all have serious reservations. I go back to that fundamental truth that there is no question, debate, difficulty or complexity. You either have the rights set out at law or you do not. That should inform all discussions and points around IP and copyright. We should have an approach that goes to the heart of this fundamental truth: it is our data. We decide, determine and choose and then, for citizens, consumers and creatives, we have a real opportunity to say positively, with a hashtag, “#OurAIFutures”.