International Relations and Defence Committee
Wednesday 21st October 2026
Skip to latest contribution
This debate has concluded
(began 1 day, 8 hours ago)
Watch LiveSkip to latest contribution
This debate has concluded
11:12
Speaker 1
-
Speaker 1
Positioning itself as a provider of public goods. For others, it can be a bit of a bind that it finds. Those will actually be for China. And there's always when when any country is positioned itself in. perhaps then being. expected by other countries to provide things that may or may not be able to provide, whether that is on the technological front or otherwise, diplomatic, etc.. So it is a key thing to keep an eye on.
It is still unfolding and.
It is still unfolding and.
11:12
11:12
11:12
Speaker 2
-
Speaker 2
Sort of the first maybe 4 or 5 years, was very focused on the development side and comparing Chinese and American capabilities in AI. And how can we sort of measure research quality and research output? How can we measure the quality of the data ecosystem, stuff like that. And then beginning around.
11:12
11:12
Speaker 2
-
Speaker 2
I shifted focus from the development side more towards the regulation and governance side. 2021 is when China started rolling out its first. meaningful regulations on AI. They have AI policy going back much further, but actual sort of binding regulations that impose requirements on companies started around then. And so I spent the last five years, you know, monitoring all aspects of Chinese , Chinese AI ecosystem with a special focus on. governance in terms of China's approach to the development and adoption of the technology. And I think this is in some ways gets to a sort of a philosophic.
11:13
11:13
Speaker 2
-
Speaker 2
Is to put a heavy. emphasis on the application of the technology. How can we use this technology to achieve our broader economic especially, but also social political. goals with. I think the core of that being, how can we use AI to sort of lever up our economy? How can we use it to climb global value chains? How can we use it to skip over the middle income trap that people are afraid we're going to fall into? And that tends to be how China looks at most new technologies, including information .
11:13
11:13
Speaker 2
-
Speaker 2
They don't don't. particularly like or not particularly interested in sort of pure software or purely financial markets. They call that the I forget what the exact translation would be, but kind of like the fake economy or the the. virtual economy. And they have a different word, the real economy for the. manufacturing of stuff for like physical services delivered in the world. And that always put a heavy emphasis on the real economy. We don't want to be like the United States who kind of abandoned that and through financialization. So when they look at a technology like AI or even in, you know, ten plus years ago, when they're looking at the mobile internet, they were thinking, how can we turn this into a tool for upgrading our economy and for achieving these sort of broader social, but especially economic goals.
And I think that really has guided a lot of their development. They have companies that are competing at the absolute frontier of AI.
And I think that really has guided a lot of their development. They have companies that are competing at the absolute frontier of AI.
11:14
11:14
Speaker 2
-
Speaker 2
And that is a really important sector. That's where the US is primarily competing. But I think if you were to, you know, give the top leaders sort of truth serum and say, what do you care the most about? Is it about the absolute frontier or is it about broad, diffuse applications of the technology? They would lean towards the application side, I guess, just briefly, in terms of how they tend to do that, this is a lot of people think of the Chinese system as very top down, and there are a lot of elements of it that are very top down.
But when it comes to how are you going to stimulate the adoption of a very diffuse.
But when it comes to how are you going to stimulate the adoption of a very diffuse.
11:15
11:15
Speaker 2
-
Speaker 2
That's not something that you can sort of perfectly plan from the top. You cannot lay out a perfect roadmap for where and when and how AI is going to be used. So in both the. 2017 National AI plan, which was kind of the first big development. document and the. 2025 AI plus plan, which is the most recent big development.
11:15
11:15
11:15
11:15
11:15
11:15
11:16
11:16
Speaker 2
-
Speaker 2
Maybe you install AI into your sort of camera networks. If you are a governor, maybe you are creating sort of a tax free AI development zone, maybe you're launching pilot projects that sort of integrate AI into public services or into factories. I think that's traditionally the way that they tend to. incentivize the. diffusion of a very.
11:16
11:16
11:16
11:16
Speaker 2
-
Speaker 2
Mostly because the local governments that are sort of charged with actually doing this work, they don't have the same type of cash floating around that they had ten years ago due to the collapse of the property bubble or the collapse of the property market. But that's, I'd say, you know. philosophically heavy focus on applications. And in terms of the mechanism, it tends to be this I think of it as like a call and response. Central government sends out a signal. You will be rewarded if you do AI well in your area, and then it's up to the localities to figure out how to do that.
I'll leave aside some questions on sort of frontier AI development or.
I'll leave aside some questions on sort of frontier AI development or.
11:17
11:17
11:17
Speaker 2
-
Speaker 2
I think I mean, there's a the first concern they have when a big information technology comes into play is, is this going to. disrupt our controls on information? And that is a heavily. ideological thing. So when China started to regulate AI, their first regulations were on recommendation algorithms, deepfakes and generative AI. And the sort of the through line in all of those was, how do we make sure that this doesn't disrupt our existing controls? You know, we knew how to control what the media says. We know how to control what goes on the internet, but how do we control what AI model says?
How do we make sure it says the right thing? So that in many ways is almost like the. prerequisite to the next step, which is the kind of go go development phase. I think of it as three kind of goes through three phases.
How do we make sure it says the right thing? So that in many ways is almost like the. prerequisite to the next step, which is the kind of go go development phase. I think of it as three kind of goes through three phases.
11:18
11:18
11:18
11:18
11:18
11:18
11:18
Lord Lamont of Lerwick
-
-
Lord Lamont of Lerwick
Lordotic has the next question. Could you just probably for for Mr. Sheehan first, but both of you may want to comment on the development of AI intelligence. I mean, how as an expert, would you say that that China sort of matches up to the to the United States? I mean, where are they really in terms of of the competition between them? And how would you characterize that competition? Is this like an arms race? Is it a Cold War? I mean,
11:19
11:19
Speaker 2
-
Speaker 2
Where you come from? So I think it's helpful to split this into. competition at the frontier of AI, who is building the biggest, strongest, most capable. models and. competition in terms of the diffusion and application of AI, which in many cases relies on having good frontier models. But it's not necessarily there's a lot of stuff that is done in robotics and. manufacturing that really doesn't need the most advanced AI. So I think if we separate those two things out, it can be a little bit cleaner. In terms of the frontier models, the nowadays, I'd say the general consensus, and I imagine Kayla talks to many of the same people that I talk to on this is that China's best models are about 6 to 9 months behind the best US models, six months to a year.
So that means essentially the model that anthropic comes out with today, in 6 to 9 months, China will come out with a model that's about equal to it. I might push it a little further back, maybe 9 to 12 months. My personal take. So that's in the frontier. That's relatively easy to quantify because we can measure the capabilities and we can kind of do head to head comparisons. On the diffusion front, it's much harder to quantify. I'd say the short version is that the US is very successful and very quick at diffusing sort of software based AI, diffusing AI into business.
So that means essentially the model that anthropic comes out with today, in 6 to 9 months, China will come out with a model that's about equal to it. I might push it a little further back, maybe 9 to 12 months. My personal take. So that's in the frontier. That's relatively easy to quantify because we can measure the capabilities and we can kind of do head to head comparisons. On the diffusion front, it's much harder to quantify. I'd say the short version is that the US is very successful and very quick at diffusing sort of software based AI, diffusing AI into business.
11:20
11:20
11:20
11:20
11:20
11:20
11:20
11:20
Speaker 2
-
Speaker 2
And having IT departments that integrate it into the systems of major. corporations in China. They don't have a long history of buying software and paying for it. And so that diffusion is slower and harder, and it's harder for the companies to make money. They partially due to what we're talking about earlier, they place their bets more on the diffusion into sort of.
11:20
11:20
11:21
11:21
Speaker 2
-
Speaker 2
I'm going to forget the last two of them. But I think China is placing its bets on that type of diffusion. Whereas we tend to place our bets when I say we you know the United States. But I think I imagine the UK is in somewhat similar boat on this tend to place our bets more on using directly integrating .
11:21
11:21
11:21
11:21
11:21
11:21
11:22
11:22
11:22
11:22
11:22
Speaker 1
-
Speaker 1
Especially on the frontier model level. Again, those metrics, I hear the same ones on. 6 to 9 months in terms of having the most. advanced frontier model. recognizing it's a. relative shortcomings in terms of competitive. advantage when it comes to compute, as well as the broader ability to. continuously develop the most advanced models. And ahead of the US, China has therefore leaned. into open, open waiting, broadly called open sourcing, but very focused on part of the models being publicly available and downloadable. And so a lot of the. industry players have really leaned into that approach over the years.
It is evolving and changing, and not every single company or developer is. continuing to open its most advanced models. And we've seen movement on that, both at the firm level as well as the country level, with proposed draft updates to export controls related to AI and other other industries. So when it comes to the as Matt was saying.
It is evolving and changing, and not every single company or developer is. continuing to open its most advanced models. And we've seen movement on that, both at the firm level as well as the country level, with proposed draft updates to export controls related to AI and other other industries. So when it comes to the as Matt was saying.
11:23
11:23
11:23
11:23
11:23
11:23
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:24
11:25
Lord Ahmad of Wimbledon
-
-
Lord Ahmad of Wimbledon
Thank you, Lord, I commend that's going to come in now. Yeah. Good morning both. Just building on the responses. You've given the previous two questions just to push that element of. philosophy and China vis a vis the US. We've seen repeatedly for many years the economic dominance of China as opposed to the US model, which actually ties in countries for a very long period of time, which means that the reliance on any renewal, any. restructuring of a debt that may be owed, is actually tied in. specifically and conditionally into the Chinese approach. So bearing in mind the kind of difference you've outlined, could you just elaborate, perhaps start with you, Matt, just where you feel that difference is arises?
And we've also had a lot of evidence on the public private expenditure, whereas the US out does China substantially. Where are we with an .
And we've also had a lot of evidence on the public private expenditure, whereas the US out does China substantially. Where are we with an .
11:26
11:26
11:26
11:26
11:26
11:26
11:26
Speaker 2
-
Speaker 2
I think started off with perhaps not that much reality behind it, but they essentially kind of stumbled into a real version of that through what Kayla described in terms of China releasing its models open weight. So well, you know, anthropic released its best model, but you can only access it by going to cloud.com. Chinese models are the entire model is uploaded to the internet. People with enough technical capability and enough computing power can just fully download the Chinese models deep sea Kimi and then they can tweak them. They can run them locally. They do not have to run.
They do not have to interact at all with Kim with a moonshot deep seek these Chinese companies. And so essentially that becomes a much, oftentimes a much cheaper way to do this for either businesses that are trying to save money or for low income countries that are not able to, you know, pay for all the subscriptions to this. And I think that shift in towards releasing the models open weight is what has given some. reality to China's claims to be a provider of global public goods on this. The sort of the flip side of that is that there is not actually a sort of a legal.
relationship between the people who are downloading and using Chinese open weight models. They're not paying for those open weight models for the most part. And they're once they're downloaded, they're entirely out of China's hands. The China does not have the ability to kind of like claw those back from people using them. And so this in many ways, from a rhetorical level, this has been very good for China in terms of giving reality to the claims that were made about China providing these global public goods.
But it also is not it's not revenue.
They do not have to interact at all with Kim with a moonshot deep seek these Chinese companies. And so essentially that becomes a much, oftentimes a much cheaper way to do this for either businesses that are trying to save money or for low income countries that are not able to, you know, pay for all the subscriptions to this. And I think that shift in towards releasing the models open weight is what has given some. reality to China's claims to be a provider of global public goods on this. The sort of the flip side of that is that there is not actually a sort of a legal.
relationship between the people who are downloading and using Chinese open weight models. They're not paying for those open weight models for the most part. And they're once they're downloaded, they're entirely out of China's hands. The China does not have the ability to kind of like claw those back from people using them. And so this in many ways, from a rhetorical level, this has been very good for China in terms of giving reality to the claims that were made about China providing these global public goods.
But it also is not it's not revenue.
11:27
11:27
Speaker 2
-
Speaker 2
It may be some knock on effects that would eventually lead to revenue, but not directly revenue generating. And it doesn't necessarily provide China with the kind of. leverage that, for example, like debt. servicing can provide a country with. It's this is going to evolve over time. So far, this the primary mode of international diffusion has been these open weight models. We're going to see this change. We're going to see, as Kayla mentioned, some of the Chinese companies are starting to close up their models, meaning not release them, open weight, releasing them in the same way that OpenAI and anthropic do.
That's going to be a major trend here, and maybe some of. China's sort of provision of these global public goods will start to look a little bit more like past generations with Belt and Road, where there are major.
That's going to be a major trend here, and maybe some of. China's sort of provision of these global public goods will start to look a little bit more like past generations with Belt and Road, where there are major.
11:28
11:28
11:28
11:28
11:28
11:29
11:29
11:29
11:29
11:29
11:29
11:29
Speaker 1
-
Speaker 1
How do these companies make enough revenue back to convince investors to continue investing, but also create revenue over time? And I think there's a lot of. experimentation in that right now from, from both types of companies. But the way that they will find success, I think, will have extreme. consequences for what lock in looks like and what. dependencies and flows of, of investment in the future look like. But to, to go back a little bit to the economic development model.
11:30
11:30
Speaker 1
-
Speaker 1
One of the things that kind of over the years China has done in and outside of the AI context is pitch other countries on its economic development model, having lifted so many out of poverty genuinely over time, but having it its unique approach and having gone to other countries, building soft power or trying to and saying, you two can go on this journey and become more of a an advanced.
11:30
11:30
11:30
11:30
11:30
11:31
11:31
11:31
11:31
11:31
11:31
Speaker 1
-
Speaker 1
And there might be more room for. intentionality at the Sei adoption of open models. It only becomes really high switching costs if you integrate it in such a way to make it hard to switch to another model when another one becomes. available in some form or another. So I think that there's a lot of security analysis needed. There's a lot of additional research needed on this, but the potential for strategic.
maneuvering to avoid real lock in dependencies is a bit more there than, say, building a whole 5G. telecommunications infrastructure or, or more.
maneuvering to avoid real lock in dependencies is a bit more there than, say, building a whole 5G. telecommunications infrastructure or, or more.
11:32
11:32
11:32
11:32
11:32
11:32
11:32
11:32
Speaker 2
-
Speaker 2
Let me I'm trying to think of balancing sort of the AI piece versus others. I think, on maybe starting with AI there. Their main goal in this is, I guess, their reputational goal, soft power goals. And then there is. economic goals for China. And so I think, you know, the their first major foray in this is the creation of the world AI cooperation Organization, Waco in Shanghai last year. And that's very much while it's still very vague what is actually happening there, and I think it's vague to the creators of the institution and vague to those countries that have signed up.
The general thrust of it is, let's help you build up your own AI. infrastructure in a way that can, you know, speed up adoption of AI in your countries. And I think there is, you know, the next step, the unstated next step of that is and we will be bringing our AI products to bear there. And we will you will be a customer there. There's it's not. entirely sort of self-interested. I do think that there is some genuine element of. believing in the power of development and believing that other countries have a right to develop.
So I don't want to pretend that it's all some master plan, but I think that's directionally that's sort of where that is pushing in terms of their engagement with the UN, because that's where they if you ask them or if you read the documents, every document about multilateral engagement on AI will say, you know, we support true. multilateralism with United Nations as the core. Why the United Nations, I think you'll you'll get very different responses from asking people in, say, the American strategic community and those in China. People in the American strategic community will be will say that this is because China knows that nothing meaningful will ever happen at the UN, or there will be no meaningful costs imposed on China, no meaningful restrictions imposed on China by way of the UN.
China would say that it's because the UN is the center of. You know, just. And what's blanking on a word? It's a little bit later in China, and I'm jet lagged. It's the area of legitimate. multilateralism that is global. I think the key, the key factor leading to that choice is China loves the UN because China always has a seat at the UN. China always has a powerful, you know, veto seat at the UN. And it is always afraid of being encircled, being excluded. When the United States and the UK were sort of jointly creating the international network of AI safety institutes in November of.
The general thrust of it is, let's help you build up your own AI. infrastructure in a way that can, you know, speed up adoption of AI in your countries. And I think there is, you know, the next step, the unstated next step of that is and we will be bringing our AI products to bear there. And we will you will be a customer there. There's it's not. entirely sort of self-interested. I do think that there is some genuine element of. believing in the power of development and believing that other countries have a right to develop.
So I don't want to pretend that it's all some master plan, but I think that's directionally that's sort of where that is pushing in terms of their engagement with the UN, because that's where they if you ask them or if you read the documents, every document about multilateral engagement on AI will say, you know, we support true. multilateralism with United Nations as the core. Why the United Nations, I think you'll you'll get very different responses from asking people in, say, the American strategic community and those in China. People in the American strategic community will be will say that this is because China knows that nothing meaningful will ever happen at the UN, or there will be no meaningful costs imposed on China, no meaningful restrictions imposed on China by way of the UN.
China would say that it's because the UN is the center of. You know, just. And what's blanking on a word? It's a little bit later in China, and I'm jet lagged. It's the area of legitimate. multilateralism that is global. I think the key, the key factor leading to that choice is China loves the UN because China always has a seat at the UN. China always has a powerful, you know, veto seat at the UN. And it is always afraid of being encircled, being excluded. When the United States and the UK were sort of jointly creating the international network of AI safety institutes in November of.
11:34
11:34
Speaker 2
-
Speaker 2
One of the big questions was, do we invite China to join this network? It was complicated because China didn't have an AI safety institute at the time. But the main issue that I was sort of going back and forth between the places trying to communicate is that this is not another example of China being. excluded from a network that is going to then be sort of turned against China to hold it down, because that's their .
11:35
11:35
11:35
Speaker 1
-
Speaker 1
Sorry sorry. Yeah. Hold on to Stephanie. Well, on the distinction that Matt made of what the US strategic community would say versus the Chinese, I agree with that. And I think my take is somewhere in the middle in that China does have a in some ways more of a say relative to other arenas in the UN. It has been quite strategic in terms of. placement across critical. positions in UN working groups and.
11:36
11:36
11:36
11:36
11:36
11:36
Speaker 1
-
Speaker 1
And partially for that reason, wanting to center the UN for AI governance discussions. It'll be very interesting to see to what. extent Waco and other initiatives at the multilateral level play out as arenas there. China is not trying to position them as a competitor to UN spaces because it's, you know, saying it supports both. And we will see like where some of these. discussions take place. I think that's something to keep an eye on of. It's not meant as a competitor to the UN. It's more intending it to be a complement. But there are there are limitless discussions that we can have at the international level.
It'll be interesting to see where the the. momentum builds potentially more .
It'll be interesting to see where the the. momentum builds potentially more .
11:37
11:37
11:37
11:37
11:37
11:37
11:37
11:38
11:38
11:38
11:38
Speaker 2
-
Speaker 2
Mr.. Shannon, do you want to kick off on that? Sure, sure. I'll kick off. Yeah, I think technical standards are the best bet there. China is a very active participant in technical standards internationally. They're very, very active on building up AI related technical standards domestically. This is one of my main areas of research right now is they sort of they started regulating AI primarily through regulatory documents with standards as kind of.
supplemental to that. And as the pace of the technology is really.
supplemental to that. And as the pace of the technology is really.
11:38
11:38
11:38
11:38
11:39
11:39
11:39
11:39
Speaker 2
-
Speaker 2
If I have the number right on that is an AI focused technical standards group that does a lot of work on sort of safety. And there's a big open question of how what kind of a role international technical standards can play in AI safety. International technical standards tend to be a very slow process, and they tend to be more about sort of.
11:39
11:39
Speaker 2
-
Speaker 2
As opposed to sort of paving the way in the future. I think China's domestic standards can be a little bit different because they can just be done quicker, and they are more directly connected into government demands. There's not it's easier to build consensus domestically on that. So I think international technical standards are a very good starting point, with the caveat that these might not move at the speed that is needed. If we are trying to work on sort of safety .
11:40
Baroness Fraser of Craigmaddie
-
-
Baroness Fraser of Craigmaddie
Related issues. Yeah, I think I'll leave it there for now. Thank you. And Miss Bloomquist, would you would you agree to focus on something like technical standards or, you know, you mentioned about China wasn't in the same place on safety as the UK and the US were. Do they need to get their own house in order domestically before they're ready to.
11:40
11:40
Speaker 1
-
Speaker 1
They would be more likely to engage. multilaterally in? Yeah, that's a great question. I think in terms of getting its own house in order, that's been something that Matt has done a lot of work on over the years, and many in the broader AI governance community of it is has been the case that a lot of Chinese AI. developers don't have as advanced of internal safety. practices as US labs, as promoted and developed by AC here and so on. And there is definite room for improvement and need for. improvement on that front, on the international.
11:41
11:41
11:41
11:41
Speaker 1
-
Speaker 1
House for a counterpart organization. But there's been a lot of effort, and there is something resembling a counterpart on that. I will say, on the the willingness to. engage on AI safety, as well as preparing for potential international crises and incidents stemming from advanced AI. I think there's an openness. There's still a lot of work to do in the practical sense, both in China and in other countries majorly as well. And so there have been efforts also supported by the UK that have gone between the US, China and the UK for starting to build these plans.
And what even would a counterpart situation in different forms of crises, because there are many forms that they could take. With AI being such a.
And what even would a counterpart situation in different forms of crises, because there are many forms that they could take. With AI being such a.
11:42
11:42
11:42
11:42
11:42
11:42
11:43
11:43
11:43
Lord Bruce of Bennachie
-
-
Lord Bruce of Bennachie
The Lord Bruce has got the next question. Thank you very much. I'm interested in the offer that China is presenting to developing countries. Namely, we've got this great technology. It's there for you to use. We'll help you use it. It's open source and you can retain the data. Very appealing. Whereas you look at.
11:43
11:43
Lord Bruce of Bennachie
-
-
Lord Bruce of Bennachie
Couldn't trust anything about them because they're only doing it for American interests. They might want to manipulate its closed source. They want to use your data. So my question is, is there a role for the UK? We've substantially dismantled our aid program. Is there a role for the UK to work with developing countries? First of all, perhaps to act as an honest broker in this. situation and also perhaps to develop some niche capacity of our own that's relevant to that? Is that something that would perhaps restore the credibility of the of the UK's engagement with developing countries, which clearly has been.
seriously undermined now, I think, Matt, you said there's an element of. self-interest as well all the time. I mean, even our aid budget, at its most.
seriously undermined now, I think, Matt, you said there's an element of. self-interest as well all the time. I mean, even our aid budget, at its most.
11:44
11:44
Lord Bruce of Bennachie
-
-
Lord Bruce of Bennachie
There was always a recognition that there was going to be benefit. There's nothing wrong with that. But the point at the end of the day, can you project say, we really, genuinely want this to be a product that benefits the whole world and doesn't become another instrument of a sort of second Cold war between the two great powers? Is there a role for the UK .
11:44
Speaker 1
-
Speaker 1
To do that? And if so, how could they do that? Perhaps? I mean, I think that's a great. characterization of the pitch that China is making to to the rest of the world as well. And in terms of the UK's role being American and actually former U.S. State Department, but choosing to be here and to do this, this work, I absolutely believe the UK has a unique role to play, and in.
11:45
11:45
11:45
11:45
Speaker 1
-
Speaker 1
And so on. And so with the AI Security Institute being such an exceptional example of. innovation on the international stage of what kind of. institutions other countries can build out to navigate what is in their own interest and need, as well as for other countries. And so using really supporting it, building out how that can be modeled and extended in expertise forms to other countries, navigating this for themselves. And when it comes to the the. infrastructure on the aid front with other countries, I think that's a great question to keep diving into of how how then could the UK bring that specialized expertise that it.
absolutely has more than almost any other country to other countries? That too, will need to. navigate many different questions across the whole of economy and society. And are these models safe to deploy in our own country? I think there is a lot of work to be done there from a government aid and support perspective, but also on a kind of services industry perspective. There are some here in London working on the AI. assurances industry, which is both kind of. public civil society as well as private sector types of many different evaluations. And plurality is needed with such an expansive technology as AI.
So really leveraging that as as a national .
absolutely has more than almost any other country to other countries? That too, will need to. navigate many different questions across the whole of economy and society. And are these models safe to deploy in our own country? I think there is a lot of work to be done there from a government aid and support perspective, but also on a kind of services industry perspective. There are some here in London working on the AI. assurances industry, which is both kind of. public civil society as well as private sector types of many different evaluations. And plurality is needed with such an expansive technology as AI.
So really leveraging that as as a national .
11:46
11:46
11:46
Speaker 2
-
Speaker 2
Thank you. What's your take on that? Yeah, very much in agreement with Kayla. I think the UK has already built up a a deserved. reputation and brand as being one of the premier centers for the testing of frontier AI models for different risks. Now that testing tends to focus on frontier risks on, you know, is this model going to, you know, go out of control, defy the defy the wishes of its creator or manager? Can it contribute to sort of uplift of normal person's ability to, to create dangerous chemical or biological weapons? That's tends to be where the, the testing work by UK AC focuses a bit more.
I could be a little bit out of date on that. That's my memory of it. But I do think one countries around the world are going to care about that, and they're going to want something of an independent voice to do that. They are, I do not think, going to just. trust the models coming out of China as being sort of safe to deploy widely across society, and they might not trust the. evaluations that come out of the US. I think the US now, it's called the KC, but formerly the AI Safety Institute in the US, they have evaluated at least one Chinese model, a deepsea model.
And when you read the kind of introduction, it was just a very. ideological document. It was, you know, laying it on thick about the role of this model in disseminating CCP propaganda, etc., etc., which is a legitimate issue and something that actually many countries will care about in a certain sense. But I don't think that those documents coming out of the US system are going to be read as sort of independent and neutral documents. And so whether it's a kind of a it's a certification.
I could be a little bit out of date on that. That's my memory of it. But I do think one countries around the world are going to care about that, and they're going to want something of an independent voice to do that. They are, I do not think, going to just. trust the models coming out of China as being sort of safe to deploy widely across society, and they might not trust the. evaluations that come out of the US. I think the US now, it's called the KC, but formerly the AI Safety Institute in the US, they have evaluated at least one Chinese model, a deepsea model.
And when you read the kind of introduction, it was just a very. ideological document. It was, you know, laying it on thick about the role of this model in disseminating CCP propaganda, etc., etc., which is a legitimate issue and something that actually many countries will care about in a certain sense. But I don't think that those documents coming out of the US system are going to be read as sort of independent and neutral documents. And so whether it's a kind of a it's a certification.
11:48
11:48
Speaker 2
-
Speaker 2
I think maybe another element of this would be first. gathering through conversation, gathering insight from developing countries about what do they actually care about when it comes to evaluating a frontier model? Are they focused on in certain ways, these frontier risks control over AI agents. Do they care more about whether or not this model is going to. infuse sort of CCP ideology into their countries? Many places like Vietnam or the Philippines would care more about that. There's a lot of different aspects, I think learning from other countries in terms of what do they care about, and then putting the UK's.
11:49
11:49
11:49
11:49
11:49
11:49
Lord Grocott
-
-
Lord Grocott
But if you think of the US on one side and China on the other as a kind of spectrum from the, you know, let let things rip as far as the US is concerned to let's regulate things as far as China's concerned. Would you I think you referred to developing countries, whatever phrase you want to use, whether it's global South or non-aligned or. They've had all sorts of different titles over the years, where would you put the balance? It's probably an unanswerable question, but I'll ask it in any question. Where would you put the the.
balance of who's winning on that argument with countries that aren't directly aligned with the US or China? Where's the balance of world opinion on on this spectrum? And .
balance of who's winning on that argument with countries that aren't directly aligned with the US or China? Where's the balance of world opinion on on this spectrum? And .
11:50
11:50
Speaker 2
-
Speaker 2
Kayla immediately after that. You know, I feel to really answer this, I need to be spending a lot more time in the global South than I do. I'm mostly in the US and China, but from meeting a lot of officials at these kind of international AI gatherings and from looking at kind of the polling on this, I think they are. overwhelmingly on the side of like, let's use this technology to upgrade our economies and to improve our livelihoods.
It's there's some irony here. You know, the US from a governmental.
It's there's some irony here. You know, the US from a governmental.
11:50
11:50
11:50
11:51
Speaker 2
-
Speaker 2
The US companies themselves are actually, you know, quite. concerned about the risks and investing a lot of time and energy into the safety side of it, even if that's not showing up in results at this point in time. But the fact that the models are just, you know, served through an API and charged for just means they're not nearly as accessible. So while the US might have a sort of a policy let it rip philosophy, the models aren't actually as accessible and diffusing as much around the world. China's pitch of like, look, we have used.
technology to lever up our economy several times over. And while we might put a lot of. regulation on the way that these models are deployed domestically in terms of speech, we want to share the models with you, and we want you to use them to lever up your own economy. I think that that message just has a lot more traction than the message either from, you know, the US government, which is American AI. dominance or the message from the frontier AI labs, which is we're going to build these exquisite products, but they're also extremely, potentially very dangerous.
And we're very worried about that. I don't think either sides of that US pitch are really landing necessarily. Well, the Chinese pitch does. At the same time, I think those people who are sort of the most I don't know what the right word is.
technology to lever up our economy several times over. And while we might put a lot of. regulation on the way that these models are deployed domestically in terms of speech, we want to share the models with you, and we want you to use them to lever up your own economy. I think that that message just has a lot more traction than the message either from, you know, the US government, which is American AI. dominance or the message from the frontier AI labs, which is we're going to build these exquisite products, but they're also extremely, potentially very dangerous.
And we're very worried about that. I don't think either sides of that US pitch are really landing necessarily. Well, the Chinese pitch does. At the same time, I think those people who are sort of the most I don't know what the right word is.
11:52
11:52
Speaker 2
-
Speaker 2
Users of the models, people around the world, including in the global South, who really want to work with the best models. If they have the ability, they are still, I think, in large part, opting for US models. That's maybe only a sliver of the sort of user base in certain countries. But even in China here, you know, people, people who work for Chinese AI companies, building frontier models still often find ways to use the US models because they think that they .
11:52
11:52
11:52
Speaker 1
-
Speaker 1
Right? Thanks, Kayla. I agree, and I think that it is a time of flux and certain trends are taking shape that will maybe over time, lead us to a clearer answer one way or another. And I think it's a time, very much, of hedging of other. countries throughout the global South and, and other middle powers of okay. As Matt was saying, like, if a country or business even or individual can afford frontier models.
11:53
11:53
11:53
11:53
11:53
11:53
11:53
Speaker 1
-
Speaker 1
Businesses, individuals out there that will opt naturally for models that are very good and very useful, but more affordable and accessible, and sometimes are often those may tend to be Chinese models. So I think that it is going to be a few years of flux on this, but maybe have different kind of tiers of advancement of economy. But this is also playing out in a time of uncertainty at the. traditional diplomatic allyship level. And I do think that pax silica will be it's kind of the US in a way, answer to Waco and other initiatives, but it is very focused on deal making at the.
infrastructure level and the effectiveness of that. We will see over time. I think there are definitely kind of deals in progress and under consideration. So I do think it's quite in the middle. Even though there is not perfect global public opinion data out there. And with that, you know, keep keep eyes on it and keep an eye on to like what is genuinely usable and .
infrastructure level and the effectiveness of that. We will see over time. I think there are definitely kind of deals in progress and under consideration. So I do think it's quite in the middle. Even though there is not perfect global public opinion data out there. And with that, you know, keep keep eyes on it and keep an eye on to like what is genuinely usable and .
11:54
11:54
11:54
11:54
11:54
11:54
Lord De Mauley
-
-
Lord De Mauley
Molly's got a question. You've both referred to China's launch of Waco while simultaneously maintaining support for the UN as the focus for multilateral cooperation on AI. You've started to address how we should understand this approach to institutional cooperation on AI, but perhaps you could expand a bit on that and on how significant the development of Waco is in the wider context of multilateral AI governance.
11:55
Speaker 2
-
Speaker 2
Perhaps, Mr. Sheehan, you'd like to start? Sure. I think that at this stage. multilateral AI initiatives are going to be most suited to most effective when it comes to sort of international diffusion of the. technology there. I think in the in the short term and maybe medium long term, they are not going to be the place where sort of.
11:55
11:55
11:55
11:55
Speaker 2
-
Speaker 2
It's not trying to sign up for any. multilateralism that constrains what it can do when it comes to developing and deploying the technology. So kind of setting aside the the governance side of things, at least for now, I think that in the. multilateral space, China is pushing aggressively, pushing this kind of diffusion narrative.
11:56
11:56
11:56
11:56
Speaker 2
-
Speaker 2
Their technological progress. So I'm very hesitant to sort of write off Waco so early on. But I do think it has kind of. sputtered out of the gate. I think they they announced it in July this year in Shanghai, and they had kind of an initial list of countries that signed up for it. And it just wasn't the most. impressive list of countries with all with all due respect to the countries, it was not the people that you would try to get to sign on to your. multilateral AI initiative from the start. I think that was due in part to skepticism on the part of a lot of countries about China's intentions and just the vagueness of what they were offering.
You know, I spoke to officials from some countries that were.
You know, I spoke to officials from some countries that were.
11:57
11:57
Speaker 2
-
Speaker 2
And they had almost no details on what it meant to sign up to this initiative. It just meant that you were signing up. And if you're not going to have something relatively concrete to offer, I mean, you can still download China's open weight models and use them. You don't need to sign up to Waco to do that. And so at this point in time, to me, Waco feels a little bit more political, I believe, and I wish I could fact check this in real time. I think one of the initial initiatives from Waco that would be.
substantive is in the area of. international standards for. interoperability of AI agents. So China's building its agents. If those agents are interacting with agents in.
substantive is in the area of. international standards for. interoperability of AI agents. So China's building its agents. If those agents are interacting with agents in.
11:57
11:57
Speaker 2
-
Speaker 2
Do they have a way to shake hands? Do they have a way to understand the identity of the other side, to understand what tools and capabilities they have access to? That's a really important question. And China is building the domestic standards on this. My guess is that these technical standards are not being built, not being designed in a lot of the other countries in Waco and China is maybe using this as a chance to export some of its technical standards, which would both would be good for China because there would be a maybe not lock in, but it would be become a lot easier to kind of continue using Chinese agents in this way.
And it would be good for the countries in a way, because they are actually have some. participation in the standard making process. And they it can be an. educating sort of process for them because they just don't have the same technical standards, bureaucracy. And so I do think there, there can be things that Waco can offer in this space. Maybe that's all still TBD, still being figured out in real time, but sort of, I don't know. It's not quite so clear as this is the Belt and Road Initiative. We're going to build a port. You're going to pay for it.
We're going to build a railroad, and it's going to connect to your .
And it would be good for the countries in a way, because they are actually have some. participation in the standard making process. And they it can be an. educating sort of process for them because they just don't have the same technical standards, bureaucracy. And so I do think there, there can be things that Waco can offer in this space. Maybe that's all still TBD, still being figured out in real time, but sort of, I don't know. It's not quite so clear as this is the Belt and Road Initiative. We're going to build a port. You're going to pay for it.
We're going to build a railroad, and it's going to connect to your .
11:58
11:58
11:58
11:58
11:58
11:58
Speaker 1
-
Speaker 1
Thank you. After that, in a different way, paralleling it to the Belt and Road Initiative, there is a little bit of a possible but TBD element of build it and they will come of create this more at the political and signaling level organization that is somewhat club like and somewhat trying to build out the political influence and clout of China as a global leader in AI. So there's there's a strong element of that. And then, you know, zooming in on what. actually is going to happen or could. happen and is actually happening under this Waco umbrella.
And in addition to technical standards, have heard at least. gestures at a couple of different things. So some of that would be educational exchanges and capacity building at more of a technical development level, in. particular with countries throughout the Global South and, and the member countries, some of them creating programs at major Chinese universities, for example. Then also there's signaling in the direction of China's going to offer some of its models to member countries. But there is a question, I think, of how is that different from what's already out on the open source platforms available for download for anyone?
And I think that that is also a bit TBD. One specific model, though, that the that China's been promoting in general and in relation to Waco is masa. It's it's weather prediction model, which I think is quite also in line with more , more digital.
And in addition to technical standards, have heard at least. gestures at a couple of different things. So some of that would be educational exchanges and capacity building at more of a technical development level, in. particular with countries throughout the Global South and, and the member countries, some of them creating programs at major Chinese universities, for example. Then also there's signaling in the direction of China's going to offer some of its models to member countries. But there is a question, I think, of how is that different from what's already out on the open source platforms available for download for anyone?
And I think that that is also a bit TBD. One specific model, though, that the that China's been promoting in general and in relation to Waco is masa. It's it's weather prediction model, which I think is quite also in line with more , more digital.
12:00
12:00
12:01
12:01
12:01
12:01
12:01
Speaker 2
-
Speaker 2
Matt, I'll come to you first. I think it would be worthwhile to engage in the sense of let's talk about this. Can you describe what this actually is? Can we figure out what it is? I don't at this point see a lot of benefit in joining Waco. I think that would primarily be, you know, China would be just very happy to sign people up to get them to lend credibility to this. But I don't see what the benefit would be to the UK or other countries at this point in time from joining.
So yeah, I guess light touch engagement to essentially kind of understand better what is going on and what is on offer,
So yeah, I guess light touch engagement to essentially kind of understand better what is going on and what is on offer,
12:02
12:02
12:02
12:02
12:02
Speaker 1
-
Speaker 1
And that AI assurance is industry. I think there. conceptually is some possible overlap, but to deploy that type of foreign. assistance well does not require doing it within a Waco umbrella. And perhaps it would be more efficient and in line with more. traditional kind of country approaches to foreign aid and development , to do it more unilaterally. But it will be worth keeping .
12:02
12:02
12:02
12:02
12:02
12:03
12:03
Speaker 2
-
Speaker 2
5000 training slots over five years or something like that. And I think that really is a potentially. meaningful offer. Most of these countries are not going to be training their own foundation models. That is a, you know. $100 billion project over the long term. But training people to be able to use and deploy the open weight models that China has already trained and is on offer, that could be a very meaningful thing and provide a meaningful for the countries and meaningful for China in terms of drawing these countries closer together. So maybe that's another area that I would.
watch in .
watch in .
12:03
12:03
12:03
Lord Alderdice
-
-
Lord Alderdice
And Lord. Alderdice has got a committee question in a way. Picking up on that, we've marked out how one of the hallmarks of China's foreign policy is an attempt to position itself as a. developmental leader and champion of the global South, or whatever we choose to call it, and that this is. influencing its approach to AI. Looking at it from the other side, do you have any sense yet as to how China is being perceived on this front by those countries to which it is offering these opportunities? Is it winning influence at the expense of the United States and the West?
Do you have any sense of that,
Do you have any sense of that,
12:05
12:05
12:05
12:05
12:05
12:05
12:05
12:05
12:05
12:05
Speaker 2
-
Speaker 2
Across the board. Thank you very much. Just a brief data point and a couple of thoughts. I mean, data point is, I spoke to one relatively senior government official, right, in the rollout of Waco, who had been asked to join. And their basic response is, we don't know what this is. And they can't answer the question when we ask them. And so we're not going to be joining. That is, from a country that is a little bit better.
positioned in this. Overall. I think for a lot of.
positioned in this. Overall. I think for a lot of.
12:06
12:06
Speaker 2
-
Speaker 2
This may be the only sort of AI. related invitation to the table that you get. You know, there are there are it's hard for people to find. I think people in every country have this, this overriding feeling of like, we need to understand this, we need to find some way to get on the train. When it comes to AI, this is this is happening one way or another, and we need to get our foot in the door, whatever metaphor you want. And if you're in one of those countries and China has invited you to be part of this admittedly vague.
organization and is saying that it will help you with some training and stuff like that. That could be it might not sound like a great offer, but it might be the best and only offer that you really get in a given year. You know, the offer from the United States is feel free to pay for our models or maybe join if you have certain critical minerals, you know, join pack silica so that we can guarantee our supply chains here. And so I do think that as, yeah, as, as, as light as it might be on details, it still might have some power.
I think the area where there might be real soft power gains is within the technical and. entrepreneurial communities in these countries who are using and deploying Chinese models. That's where it's like a very. tangible thing that's being provided that they would not have access to otherwise, that it really empowers them to do different things. And so I think that that's having that power. But that, again, doesn't come that .
organization and is saying that it will help you with some training and stuff like that. That could be it might not sound like a great offer, but it might be the best and only offer that you really get in a given year. You know, the offer from the United States is feel free to pay for our models or maybe join if you have certain critical minerals, you know, join pack silica so that we can guarantee our supply chains here. And so I do think that as, yeah, as, as, as light as it might be on details, it still might have some power.
I think the area where there might be real soft power gains is within the technical and. entrepreneurial communities in these countries who are using and deploying Chinese models. That's where it's like a very. tangible thing that's being provided that they would not have access to otherwise, that it really empowers them to do different things. And so I think that that's having that power. But that, again, doesn't come that .
12:07
12:07
12:07
12:07
12:08
Lord Lamont of Lerwick
-
-
Lord Lamont of Lerwick
What the Chinese are doing. Like Lord Lamont has a question. It could be argued that China is endeavoring to fill the space that's been left by the United States, given its attitude to competition with China. Although I have to say that the witnesses we've had to this committee have generally denied that China wants to replace the United States in international. institutions and prefers preeminence or a leading role simply because of the cost and inherent cautiousness.
But maybe with AI it's different. It'd be interesting. That's the first question. The second question is, in general.
But maybe with AI it's different. It'd be interesting. That's the first question. The second question is, in general.
12:08
12:08
12:08
12:08
12:08
12:08
12:09
12:09
Speaker 2
-
Speaker 2
Intrusive towards China or intrusive towards other countries. The sort of number one principle that China has carried over from its view of internet international internet governance into its view of AI governance, is respecting the sovereignty of countries when it comes to their decisions about how they want to manage the technology domestically. So in the internet era, this was internet sovereignty, cyber sovereignty, which, you know, it began very much from the China defending its own position when China was one of the few countries that had really. meaningful controls on the internet. It described those as a sovereignty issue.
And it said that every country has the right to impose these. You know, that doesn't mean China respects everybody's sovereignty in cyberspace. Well, that's the principle that they carry over. And I think if they were to lead, take on some leading role or guiding role in global AI governance, I think central to that would be prioritizing sovereignty and saying that it is not the job of a global. regulator to come in and say, you know, you cannot do censorship that way. You cannot regulate your models that way, that the core of activity has to be at the national level.
So I think that's maybe their, their that'd be a starting point for a global governance regime. And then from there, they would probably move into focusing on AI for.
And it said that every country has the right to impose these. You know, that doesn't mean China respects everybody's sovereignty in cyberspace. Well, that's the principle that they carry over. And I think if they were to lead, take on some leading role or guiding role in global AI governance, I think central to that would be prioritizing sovereignty and saying that it is not the job of a global. regulator to come in and say, you know, you cannot do censorship that way. You cannot regulate your models that way, that the core of activity has to be at the national level.
So I think that's maybe their, their that'd be a starting point for a global governance regime. And then from there, they would probably move into focusing on AI for.
12:10
12:10
Speaker 2
-
Speaker 2
Of systems. Can we make sure that our systems can be deployed globally and that, you know, for example, the agents are able to connect with each other in the right way. I, I think questions around. frontier AI safety would in China's mind, be left to a domestic issue within China, a domestic issue within the US, and something that maybe those two countries are going to talk to each other about. But I don't think they would put .
12:10
12:10
12:11
12:11
12:11
12:11
Speaker 2
-
Speaker 2
But it does not want to take on the. responsibilities of being a, you know, international or even global security provider. It doesn't, you know, partly because of the sovereignty issues it describes, but it just it recognizes that that's quite burdensome. And maybe China's most successful era of. development was one in which the United States still played a role in maintaining a variety of global norms and global. institutions and security guarantees for other countries. And China was able to sort of develop on its own terms and leverage that international system, leverage the World Trade. Organization for its own development.
So I think I'd agree that they're not looking for this kind of. displacement if it means.
So I think I'd agree that they're not looking for this kind of. displacement if it means.
12:12
12:12
12:12
12:12
12:12
12:12
12:12
12:12
12:12
12:12
12:12
Speaker 2
-
Speaker 2
Definitely going back, as Matt mentioned, to internet sovereignty. And now it's making appearances here in the multilateral and global AI governance space, which is quite, quite interesting. And I think that does connect very directly to the space left by the US. China has been very. publicly criticizing the US on its approach to global affairs and.
12:13
12:13
Speaker 2
-
Speaker 2
Tied especially to the sovereignty and. domestic ways of. governing point. For a long time and then leading into the AI era. So I think that will continue to be it's part of its selling point. Is AI different from previous times of. I also agree that China has a not forward leaning and. hesitant relationship to being, or not necessarily aspiring to be a global hegemon in the way that the US was or others have been in the past.
I think that is is quite strongly upheld by a number of.
I think that is is quite strongly upheld by a number of.
12:14
12:14
12:14
12:14
Speaker 2
-
Speaker 2
As a leader of the global South, it's kind of almost. accidentally positioning. itself as being in this role of responsibility. And I think there's. domestic in terms of the expert level and kind of advisor level experts on AI governance. There has been hesitation of around Waco and around other. initiatives to what the. consequences of, of this positioning as kind of the. responsible leader and convener of Waco or other initiatives are.
because even if you are narrow in the. offerings within Waco or other.
because even if you are narrow in the. offerings within Waco or other.
12:15
12:15
Speaker 2
-
Speaker 2
Other countries might come to the. table and come with their membership now expecting certain benefits that China may not even have offered in the first place. And I think in some conversations with with people who were involved in the shaping of Waco at the more academic level, they were. internally seeming quite concerned. about this, this positioning and its. consequences thereon. So I think it'll it'll remain a complicated relationship. And some of it depends on how other. countries force China into that position. Now that it has kind of claimed great power status, even if it's not trying to .
12:15
12:15
Lord Houghton of Richmond
-
-
Lord Houghton of Richmond
Lord Hutton has a question. Well, to be honest, my question was answered by the the the answer you gave to Lord Bruce. If I could go somewhere else with the question. And it's really our committee is both international affairs, international relations and defence, and hasn't really been a hard. defence question about international.
12:16
12:16
Lord Houghton of Richmond
-
-
Lord Houghton of Richmond
And particularly in the context of artificial intelligence and autonomous weapon systems. And I know that. particularly in the Carnegie Endowment for International Peace, you do sort of you are international advocates for peaceful outcomes. And I know that you. advocate for the retention of meaningful human control within the use of. artificial weapon systems, but in many.
12:16
12:16
12:17
12:17
12:17
12:17
12:17
12:17
Lord Houghton of Richmond
-
-
Lord Houghton of Richmond
Natural military advantage. So are you sort of, you know, tilting at windmills in a way, or is there some some else within the regulatory body other than straightforward safety that you're looking at here? And if I come to you that, Matt, first, because you represent an endowment that is meant to be doing some of this stuff. And then, Kayla,
12:17
Speaker 1
-
Speaker 1
You might have a supplementary, supplementary. I'm afraid I don't really have a great answer for this. People at Carnegie do do this work and do it very well. My colleague Steve Feldstein, I think, is kind of at the forefront of many of these issues, but I really don't focus on military applications of AI. My gut instinct is that what you said at the end there about tilting at windmills? This is seen as such a. potentially the source of so much potential advantage on the. battlefield that it is quite hard to imagine countries. really showing restraint in this.
But that's that's an amateur.
But that's that's an amateur.
12:18
12:18
Speaker 2
-
Speaker 2
Evidence research. But is there anything that you would you would offer in. This isn't my area of speciality either. However, my colleague Thomas, what's in the back here might have some helpful thoughts after the. The supplementary point I would add is especially in terms of autonomous weapons and and figuring these things out, I think coalitions of middle powers are. specifically relevant here and there basically would need to be a coalition of middle powers with the.
12:18
12:18
12:19
12:19
Lord Robertson of Port Ellen
-
-
Lord Robertson of Port Ellen
Of these new capabilities. Okay. We will we will pursue this at some later evidence sessions. But the rest of what you have said today has been profoundly useful and. stimulating as well. And thank you so much for giving us your expertise, which has. clarified a lot of the a lot of the issues that we will have to, to deal with as well. So we thank you very much for .
12:19
12:19