International Relations and Defence Committee
Wednesday 14th October 2026
Skip to latest contribution
This debate has concluded
(began 1 day, 14 hours ago)
Watch LiveSkip to latest contribution
This debate has concluded
11:27
11:27
Speaker 1
-
Speaker 1
Well, one thing I would add is that I don't see anything else driving. If we want to call it multilateralism today from the US government as it stands. But I do see lots of calls to recognize that this is a global technology and as a consequence, will require global responses and approaches. I just don't see it in the US government. But we all know that there going to be more crises, first and foremost, as as individuals, institutions and the agents themselves that are being created, exploit, find and exploit cyber vulnerabilities in pursuit of their goals and lead to consequences that are disruptive.
The ability that's not just something that we're going to see from American models or from Chinese models, that that's a danger that will come from literally everywhere. And, and the only way to effectively respond to that is to have a governance regime that has some global capabilities. Now, the initial response of this, you've seen from people like Demis Hassabis, formerly DeepMind, now Google AI and and from Dario Amodei at Anthropic talking about a a Finra style model for artificial intelligence, in a sense, a corporate and industry run, technocratic, nominally independent oversight board that identifies threats, communicates them, and tries to contain them.
A significant step beyond that would be the Technology Stability Board that Mustafa Suleiman and I have been talking about for a year, a few years now. He runs AI at Microsoft today, but that Mark Carney from Canada has recently promoted, with some support from a number of other so-called middle powers. And that would take this idea and multilateral ize it with a technocratic oversight with the same intentions in mind. It just would not be limited to the private sector.
Relevant players in the United States. And and then more broadly, you can start talking about ministries of AI that are themselves acting as independent from from governments because they need to have orientation to safeguard the stability of AI systems, which we all need to breathe and to conduct the economy without being politicized by their governments. In other words, you know, much as I suggested that the present way that the US government is thinking about AI is through a Cold War lens and an arms race lens, where AI default becomes the next nuclear weapon, just much faster and much more dangerous.
I'm suggesting that the appropriate analogy for AI is the financial marketplace. And the interesting thing about the financial marketplace, despite all of the dysfunction in global governance today and the lack thereof, is that everybody understands that we need the financial marketplace to function number one. Number two, the threats to the financial marketplace can come from anywhere and can quickly become systemic. So interestingly, whether you are the Bank of England or you're the ECB or the fed or the PBoC in China, very different systems politically.
And yet all understand that threats need to be surfaced, communicated and , and, and contained when they when they occur. And all therefore have to exhibit a level of independence from the political cycles and divisions inside their countries. Now, again, we are nowhere close to that orientation in the US government and certainly in the Trump administration today. But but I do think I mean, I can think of a number of senators today that are very interested in in discussing that orientation.
There are CEOs that are interested in discussing that orientation, and there are other government leaders that are interested. So if we're thinking about multilateralism from the US, especially on AI, I would I would hesitate, I guess I would I would urge you not just to think about where President Trump is today as some kind of fixed strategic point. And I dare not use Trump and strategic in the same sentence, but but rather about the fact that this is a moving target at speed.
And I suspect that US engagement in in potential multilateral and global solutions is likely to look very, very different even before Trump leaves power.
The ability that's not just something that we're going to see from American models or from Chinese models, that that's a danger that will come from literally everywhere. And, and the only way to effectively respond to that is to have a governance regime that has some global capabilities. Now, the initial response of this, you've seen from people like Demis Hassabis, formerly DeepMind, now Google AI and and from Dario Amodei at Anthropic talking about a a Finra style model for artificial intelligence, in a sense, a corporate and industry run, technocratic, nominally independent oversight board that identifies threats, communicates them, and tries to contain them.
A significant step beyond that would be the Technology Stability Board that Mustafa Suleiman and I have been talking about for a year, a few years now. He runs AI at Microsoft today, but that Mark Carney from Canada has recently promoted, with some support from a number of other so-called middle powers. And that would take this idea and multilateral ize it with a technocratic oversight with the same intentions in mind. It just would not be limited to the private sector.
Relevant players in the United States. And and then more broadly, you can start talking about ministries of AI that are themselves acting as independent from from governments because they need to have orientation to safeguard the stability of AI systems, which we all need to breathe and to conduct the economy without being politicized by their governments. In other words, you know, much as I suggested that the present way that the US government is thinking about AI is through a Cold War lens and an arms race lens, where AI default becomes the next nuclear weapon, just much faster and much more dangerous.
I'm suggesting that the appropriate analogy for AI is the financial marketplace. And the interesting thing about the financial marketplace, despite all of the dysfunction in global governance today and the lack thereof, is that everybody understands that we need the financial marketplace to function number one. Number two, the threats to the financial marketplace can come from anywhere and can quickly become systemic. So interestingly, whether you are the Bank of England or you're the ECB or the fed or the PBoC in China, very different systems politically.
And yet all understand that threats need to be surfaced, communicated and , and, and contained when they when they occur. And all therefore have to exhibit a level of independence from the political cycles and divisions inside their countries. Now, again, we are nowhere close to that orientation in the US government and certainly in the Trump administration today. But but I do think I mean, I can think of a number of senators today that are very interested in in discussing that orientation.
There are CEOs that are interested in discussing that orientation, and there are other government leaders that are interested. So if we're thinking about multilateralism from the US, especially on AI, I would I would hesitate, I guess I would I would urge you not just to think about where President Trump is today as some kind of fixed strategic point. And I dare not use Trump and strategic in the same sentence, but but rather about the fact that this is a moving target at speed.
And I suspect that US engagement in in potential multilateral and global solutions is likely to look very, very different even before Trump leaves power.
11:33
Lord Lamont of Lerwick
-
-
Lord Lamont of Lerwick
Thank you very much. Any of you would like to comment further on that? And perhaps the analogy with the regulation of financial services, I mean, regulation of financial services, we've observed how regulators have huge difficulty in keeping up to date with the developments in the market. One would have thought there would be even more difficult in the case of AI.
11:33
11:42
Lord Ahmad of Wimbledon
-
-
Lord Ahmad of Wimbledon
You know, the kill button. Increasingly every day you hear new reports of how AI may be used in a way to through attacks. We saw, for example, yesterday in France with the cyber attack on supermarket, where the frozen food went to heated proportions and they had to dispose. It opens up the floodgates. And even if the private sector regulates from a Western perspective, from an American perspective, does that not still leave the big question of what China will be doing in other areas where it has influence?
11:43
Speaker 2
-
Speaker 2
Yeah, so let me address those consecutively. So first, on how seriously we should take these calls and efforts by the private sector. I think we should take them pretty seriously. And the reason I think that is because if you look at something like the the recent super intelligence accord and the voluntary commitments that companies agreed to when they signed that document, this builds on a long list of voluntary commitments that AI companies have made, including with the prior administration, including in various international contexts around elections, for example, at the Munich Conference around safety frameworks, at the Seoul AI summit and the other summits in the AI summit series.
What has happened is that they have sort of developed industry practices. They've committed to them publicly. They've become de facto industry standards. And then those standards, in turn have been codified into law. So, for example, some of the state frontier AI laws you see in the United States, laws like those passed in California, New York and Illinois build on these sort of de facto standards that AI companies have developed as they have figured out how to or figured out their safety approaches to their technology.
And as that has developed and improved over time. And so this this recent accord, I think, could very well see the same pattern where it is a voluntary agreement, it is implemented by labs and then eventually becomes the scaffolding for some sort of regulatory legislative framework that could happen at the US state level. That could happen at the US national level, though, though it's hard to see how that gets around a veto from the president.
But it could also sort of anchor international conversations. On the. The other question, sorry, can you repeat the the last part? If the
What has happened is that they have sort of developed industry practices. They've committed to them publicly. They've become de facto industry standards. And then those standards, in turn have been codified into law. So, for example, some of the state frontier AI laws you see in the United States, laws like those passed in California, New York and Illinois build on these sort of de facto standards that AI companies have developed as they have figured out how to or figured out their safety approaches to their technology.
And as that has developed and improved over time. And so this this recent accord, I think, could very well see the same pattern where it is a voluntary agreement, it is implemented by labs and then eventually becomes the scaffolding for some sort of regulatory legislative framework that could happen at the US state level. That could happen at the US national level, though, though it's hard to see how that gets around a veto from the president.
But it could also sort of anchor international conversations. On the. The other question, sorry, can you repeat the the last part? If the
11:45
Speaker 2
-
Speaker 2
Yeah. So so my observation there is that when I talk to employees at AI companies, they a lot of them are alarmed by the pace at which technology is developing. They would like it to slow down. You see, you see these public letters, but the companies don't feel like they can slow down. They feel like they're in a prisoner's dilemma. And that that prisoner's dilemma operates at two levels. So it operates first within the United States.
There is basically the rationale that if I work in an AI company, an AI company unilaterally decides to slow down. That won't matter because my competitor companies won't. And so that is problematic. And coupled with that, there is a thought that as technology marches upward that increases the risk surface, it also increases the number of tools that are available to address some of those risks. Maybe not at the same pace, but there is a thought that, especially if my competitors continue to develop capabilities, then we should also develop capabilities.
Because I think of myself as working at the more responsible company. And so I will be able to leverage those tools to address risk better than my competitors would. Then on top of that, you have this prisoner's dilemma occurring between the US and China, where even if you get some sort of collective action to slow down technology development in the US, that won't matter, because that won't apply to Chinese companies. They're already by by, by most accounts, narrowing the gap between the best Chinese models and the best US models.
And so any pause on the US side would allow China to further narrow or even eliminate that gap. And so the only solution I see for this is for some sort of narrow agreement between the US and China. That wouldn't harm the ability for companies to compete in the open market, right? For companies to, to say, you know, advocate for their products and say our products are better and you should choose them over competitors.
But that would sort of apply some sort of a break or slowdown or regulatory system that would apply equally to the US and Chinese industry. That that's very hard to do for, for lots of reasons, figuring out the mechanisms for how to do this in a way that doesn't harm the overall ability for these countries to compete is one of them. But I think the harder issue is that how can you build in the verification mechanisms that would allow the US and China to trust what the other country is saying? That that seems incredibly difficult, but that is the the only long term solution I see here to to address this anxiety about how fast AI technology is developing faster than even a lot of people in AI companies are comfortable with.
There is basically the rationale that if I work in an AI company, an AI company unilaterally decides to slow down. That won't matter because my competitor companies won't. And so that is problematic. And coupled with that, there is a thought that as technology marches upward that increases the risk surface, it also increases the number of tools that are available to address some of those risks. Maybe not at the same pace, but there is a thought that, especially if my competitors continue to develop capabilities, then we should also develop capabilities.
Because I think of myself as working at the more responsible company. And so I will be able to leverage those tools to address risk better than my competitors would. Then on top of that, you have this prisoner's dilemma occurring between the US and China, where even if you get some sort of collective action to slow down technology development in the US, that won't matter, because that won't apply to Chinese companies. They're already by by, by most accounts, narrowing the gap between the best Chinese models and the best US models.
And so any pause on the US side would allow China to further narrow or even eliminate that gap. And so the only solution I see for this is for some sort of narrow agreement between the US and China. That wouldn't harm the ability for companies to compete in the open market, right? For companies to, to say, you know, advocate for their products and say our products are better and you should choose them over competitors.
But that would sort of apply some sort of a break or slowdown or regulatory system that would apply equally to the US and Chinese industry. That that's very hard to do for, for lots of reasons, figuring out the mechanisms for how to do this in a way that doesn't harm the overall ability for these countries to compete is one of them. But I think the harder issue is that how can you build in the verification mechanisms that would allow the US and China to trust what the other country is saying? That that seems incredibly difficult, but that is the the only long term solution I see here to to address this anxiety about how fast AI technology is developing faster than even a lot of people in AI companies are comfortable with.
11:48
11:48
Lord Grocott
-
-
Lord Grocott
Yes. I mean, you've touched on this quite a bit already, but if I can folk, it's about US politics and a a party politics and AI. I mean, to what extent is there a divide between the two parties? I understand the two extreme views about the benefits or risks of AI, but are we moving towards a position where there would be a say in 2028, a party, a recognisable party, political divide on where the parties stand on on that spectrum as to regulation or no regulation, to put it crudely.
And and if I can just add a rider to that, it would be this more personal question, really, for people who are not specialists in this subject, which includes me. To what extent do you. Where are you on this spectrum? I mean, is this something about which, as some recent commentaries have suggested, the development of AI have been something about which we should be intensely anxious and worried and determined to deal with and tackle? Or can we take a slightly more relaxed view of this and, and see, there are there are benefits and we've got to watch it, but let's not get too overexcited about it.
So two questions there. Really.
And and if I can just add a rider to that, it would be this more personal question, really, for people who are not specialists in this subject, which includes me. To what extent do you. Where are you on this spectrum? I mean, is this something about which, as some recent commentaries have suggested, the development of AI have been something about which we should be intensely anxious and worried and determined to deal with and tackle? Or can we take a slightly more relaxed view of this and, and see, there are there are benefits and we've got to watch it, but let's not get too overexcited about it.
So two questions there. Really.
11:50
11:50
Speaker 1
-
Speaker 1
in my life professionally seen an issue where the gap between capabilities and governance has been so large, so profound and so impactful. There is no issue that I think we need to concern ourselves with more, that we are nowhere close to where we need to be than in the appropriate governance of AI. One thing we haven't talked about is the fact that the US economy would be in a very deep recession if it wasn't for the success of AI at the frontier, and the incredible amount of money that is that is attached to that.
And yet you've got not just the companies themselves, but you have this massive internal backlash. So, I mean, even if you didn't believe that any of the things that we were talking about were likely to transpire, simply the vulnerabilities of the US economy and all around the world that rely on it to this goose that lays the golden eggs , that by itself would make you take this incredibly seriously. Because this is a train that's moving with great speed, and nobody knows what happens if it suddenly stops.
So there's that on the on the party side. I mean, of course, so much depends on what kind of candidates you end up getting in 2028, because there's a wide variety of where they would stand on this issue, but there's some clear party divides. Elon Musk is now the single biggest funder of the Republican Party, unprecedented amounts of personal money that he is devoting into these races, and I fully expect him he'll do that again in 2028.
And he has a very specific view of artificial intelligence and of a hands off superintelligence is coming. Humans are going to be much less powerful and much less intelligent than these AI bots and agents very soon. And we should embrace that. And yeah, there's a 10% chance or whatever that we're going to destroy humanity, but doesn't matter. We need to keep going and, and embrace our integration of human society with these capabilities.
That's, that's a very specific response. It's not a response you'll find anywhere in the Democratic Party. Meanwhile, you're going to see the Democrats who will win in a few weeks on the back of not their own great ideas, but rather comparatively negative backlash to issues of affordability, corruption and incompetence of the Trump administration. You're going to see them leaning in to this idea that behind Trump and the guy that promised to drain the swamp but hasn't, are all of these billionaires and multibillionaires that were standing behind him on the dais of Inauguration Day and are putting money in and pumping money in to him and his family, and they're not benefiting you.
So at the very minimum, you're going to see a very significant gap over the next two years between Democrats and Republicans on that. Now, having said that, when I mentioned at the opening of this conversation that there is a very significant backlash against data centers that's being felt in Mississippi and in Michigan. So red states and blue states are facing that, and those leaders are having to respond and say, yeah, yeah, yeah, yeah.
We want to make sure that your electricity prices are not going up. And that's a real concern. We need to make sure that your water supply is not going away. That's not a real concern. We need to make sure that zoning regulations are being followed and that we're not, you know, sort of creating massive growth in inequality because of the haves and have nots that come on the back of all of this investment.
I think the Republican Party will be more oriented towards box checking exercises to slow down. But then after elections, announce a bunch more of these new investments that grow that that that use growth. And I think Democrats will be more focused on more , if I can use the term nanny state orientation of more intrusive regulation and the behaviors of the AI companies needing to be limited in terms of, you know, what they do with education and what they do with jobs and what they do with things like, you know, occupational therapy, even all other things.
So I do think that a gap is coming, but on that front, it's going to be more individual candidate driven as opposed to the broader, more yawning gap that I mentioned in the outset.
And yet you've got not just the companies themselves, but you have this massive internal backlash. So, I mean, even if you didn't believe that any of the things that we were talking about were likely to transpire, simply the vulnerabilities of the US economy and all around the world that rely on it to this goose that lays the golden eggs , that by itself would make you take this incredibly seriously. Because this is a train that's moving with great speed, and nobody knows what happens if it suddenly stops.
So there's that on the on the party side. I mean, of course, so much depends on what kind of candidates you end up getting in 2028, because there's a wide variety of where they would stand on this issue, but there's some clear party divides. Elon Musk is now the single biggest funder of the Republican Party, unprecedented amounts of personal money that he is devoting into these races, and I fully expect him he'll do that again in 2028.
And he has a very specific view of artificial intelligence and of a hands off superintelligence is coming. Humans are going to be much less powerful and much less intelligent than these AI bots and agents very soon. And we should embrace that. And yeah, there's a 10% chance or whatever that we're going to destroy humanity, but doesn't matter. We need to keep going and, and embrace our integration of human society with these capabilities.
That's, that's a very specific response. It's not a response you'll find anywhere in the Democratic Party. Meanwhile, you're going to see the Democrats who will win in a few weeks on the back of not their own great ideas, but rather comparatively negative backlash to issues of affordability, corruption and incompetence of the Trump administration. You're going to see them leaning in to this idea that behind Trump and the guy that promised to drain the swamp but hasn't, are all of these billionaires and multibillionaires that were standing behind him on the dais of Inauguration Day and are putting money in and pumping money in to him and his family, and they're not benefiting you.
So at the very minimum, you're going to see a very significant gap over the next two years between Democrats and Republicans on that. Now, having said that, when I mentioned at the opening of this conversation that there is a very significant backlash against data centers that's being felt in Mississippi and in Michigan. So red states and blue states are facing that, and those leaders are having to respond and say, yeah, yeah, yeah, yeah.
We want to make sure that your electricity prices are not going up. And that's a real concern. We need to make sure that your water supply is not going away. That's not a real concern. We need to make sure that zoning regulations are being followed and that we're not, you know, sort of creating massive growth in inequality because of the haves and have nots that come on the back of all of this investment.
I think the Republican Party will be more oriented towards box checking exercises to slow down. But then after elections, announce a bunch more of these new investments that grow that that that use growth. And I think Democrats will be more focused on more , if I can use the term nanny state orientation of more intrusive regulation and the behaviors of the AI companies needing to be limited in terms of, you know, what they do with education and what they do with jobs and what they do with things like, you know, occupational therapy, even all other things.
So I do think that a gap is coming, but on that front, it's going to be more individual candidate driven as opposed to the broader, more yawning gap that I mentioned in the outset.
11:55
11:55
Speaker 2
-
Speaker 2
Yeah. So so first on your question about how seriously we should take risk overall, I my approach is very pragmatic. So in a lot of ways, I think it it doesn't matter exactly what your assessment of the probability of something extremely serious or catastrophic is. Even if you think there's a small probability given the potential scope here, it makes sense to take a precautionary approach that is better to be prepared to put in the regulatory and institutional infrastructure in place, to be able to manage this and not have to use it, than be in a situation where you need something and haven't prepared adequately.
And so, so that's really how I think about it. And so I guess the answer would be we should, we should prepare for it seriously, because it is better to have done that preparation and not need it than vice versa. On the question about US politics, my assessment is that. You know, we have these midterm elections. It will provide a very useful data point on how AI sentiment will will play into US politics over the next few years.
My best guess for what will happen is that AI issues, particularly related to data centers, will be a significant issue in a few races. It'll be sort of an underlying simmering issue throughout the country. It will. Sort of have a marginal over impact, overall impact on the election outcomes compared to some of the other macroeconomic factors or the the dissatisfaction with the Trump administration. But by the the next election cycle for a US president, I find it hard to envision that AI will not be a major issue, that it will not be something that candidates have to to campaign on explicitly, and that by 2028, unless there is some significant action to regulate AI that I don't see happening in the current political environment, the sentiment around AI will be worse in the US, I think.
I think it's already hard to see how you can reverse that, especially as anxiety around job losses increase. But but if but part of the solution here is to regulate and answer what the public is saying about their concerns about the technology, their desire for there to be more oversight of AI technology. And if that doesn't happen, then that sentiment will be worse. And in that kind of environment, I find it hard to in any Democratic candidate in the United States not running on some sort of platform of more accountability for the tech industry, some some relatively robust accountability, as well as potentially some big ideas about around social safety nets and dealing with AI driven job loss.
I think that. Republican candidates will maybe be somewhat responsive to that pressure, but their their natural inclination is not to regulate. They will continue to rely on self-regulation from the from the industry. They will continue to be very concerned about competition with China. And so , so I, I don't see that gap going away by, by 2028, even if as I as I expect, sentiment around AI continues to become more negative in the next two and a half years.
And so, so that's really how I think about it. And so I guess the answer would be we should, we should prepare for it seriously, because it is better to have done that preparation and not need it than vice versa. On the question about US politics, my assessment is that. You know, we have these midterm elections. It will provide a very useful data point on how AI sentiment will will play into US politics over the next few years.
My best guess for what will happen is that AI issues, particularly related to data centers, will be a significant issue in a few races. It'll be sort of an underlying simmering issue throughout the country. It will. Sort of have a marginal over impact, overall impact on the election outcomes compared to some of the other macroeconomic factors or the the dissatisfaction with the Trump administration. But by the the next election cycle for a US president, I find it hard to envision that AI will not be a major issue, that it will not be something that candidates have to to campaign on explicitly, and that by 2028, unless there is some significant action to regulate AI that I don't see happening in the current political environment, the sentiment around AI will be worse in the US, I think.
I think it's already hard to see how you can reverse that, especially as anxiety around job losses increase. But but if but part of the solution here is to regulate and answer what the public is saying about their concerns about the technology, their desire for there to be more oversight of AI technology. And if that doesn't happen, then that sentiment will be worse. And in that kind of environment, I find it hard to in any Democratic candidate in the United States not running on some sort of platform of more accountability for the tech industry, some some relatively robust accountability, as well as potentially some big ideas about around social safety nets and dealing with AI driven job loss.
I think that. Republican candidates will maybe be somewhat responsive to that pressure, but their their natural inclination is not to regulate. They will continue to rely on self-regulation from the from the industry. They will continue to be very concerned about competition with China. And so , so I, I don't see that gap going away by, by 2028, even if as I as I expect, sentiment around AI continues to become more negative in the next two and a half years.
11:59
11:59
11:59
Speaker 1
-
Speaker 1
Just that, reiterating how incredibly fast moving this is, and also the gap between what the people inside the AI companies not just believe, but also their knowledge set and, and the utter absence of that inside the American political leadership. And I'm not just talking about the white House. I'm also talking about Democrats and Republicans in House and Senate. They are playing catch up. The greatest concern here for me is that the regulatory environment and the eventual emergence of governance is going to be subverted to whatever the big crisis happens to be.
And then when you end up with is a nine over 11 reaction or a shoe bomber sort of reaction where everyone saw, oh my God, this guy was going to put explosives in his shoes. And so now we all have to take our shoes off, going through TSA going on an airport, but we're not actually addressing the broader issues at hand. I mean, I'll look and I agree strongly about the concerns that are growing about social safety net and job displacement.
We're concerned about what these AI models have the capacity to do, and the lack of awareness of how they function at the frontier. All of that is real. And yet we all know that governments are not just slow moving, but it's very easy to get everyone to suddenly stand up and rush to the crisis that you just had. And, and so it's really important to have conversations like this where we're starting to do much more to anticipate the kinds of governance that we will all need strategically and long term so that when the next crisis occurs and it will occur, the cost, real resources, and everyone's jumping on it, we can take advantage of that.
We can use that crisis as an opportunity , as opposed to having our hair on fire, especially in an environment where American political leaders don't have a lot of knowledge, don't have a lot of capabilities, and are sort of relying on the CEOs to do the action. The other final thing I would say is that do not underestimate just how busy and fast moving these companies are that they are. I do believe that they're not evil.
I think that they are mostly well intentioned. I think that they really would like to be seen as doing the right thing and building products that are transforming society in the right way. But the incredible pressure to get to the next model, the all consuming. If we don't do that, we will die. And we, our families will will not have the opportunities we thought of. And everything we're trying to build is instead going to be in someone else's pocket.
And maybe people we don't trust at all that that is overwhelmingly consuming what they're doing. And therefore, and if that's the case, even if they can have some conversations about the things that we're discussing today, they are in no position to drive governance. So we've got this environment where the technology companies are acting as functional sovereigns in the governance space of artificial intelligence, and yet they are utterly unprepared to do so. And there is nothing that will happen in the next year or two years that is going to make them better prepared than the governments actually are, despite the talent gap.
So we can't abdicate the role that the governments have. Even as AI looks like an increasingly techno polar space as opposed to unipolar or multipolar. So again, I thank you very much for inviting me. I'm sorry I'm running to an airport right now, but I'm sure we'll have a chance to chat again.
And then when you end up with is a nine over 11 reaction or a shoe bomber sort of reaction where everyone saw, oh my God, this guy was going to put explosives in his shoes. And so now we all have to take our shoes off, going through TSA going on an airport, but we're not actually addressing the broader issues at hand. I mean, I'll look and I agree strongly about the concerns that are growing about social safety net and job displacement.
We're concerned about what these AI models have the capacity to do, and the lack of awareness of how they function at the frontier. All of that is real. And yet we all know that governments are not just slow moving, but it's very easy to get everyone to suddenly stand up and rush to the crisis that you just had. And, and so it's really important to have conversations like this where we're starting to do much more to anticipate the kinds of governance that we will all need strategically and long term so that when the next crisis occurs and it will occur, the cost, real resources, and everyone's jumping on it, we can take advantage of that.
We can use that crisis as an opportunity , as opposed to having our hair on fire, especially in an environment where American political leaders don't have a lot of knowledge, don't have a lot of capabilities, and are sort of relying on the CEOs to do the action. The other final thing I would say is that do not underestimate just how busy and fast moving these companies are that they are. I do believe that they're not evil.
I think that they are mostly well intentioned. I think that they really would like to be seen as doing the right thing and building products that are transforming society in the right way. But the incredible pressure to get to the next model, the all consuming. If we don't do that, we will die. And we, our families will will not have the opportunities we thought of. And everything we're trying to build is instead going to be in someone else's pocket.
And maybe people we don't trust at all that that is overwhelmingly consuming what they're doing. And therefore, and if that's the case, even if they can have some conversations about the things that we're discussing today, they are in no position to drive governance. So we've got this environment where the technology companies are acting as functional sovereigns in the governance space of artificial intelligence, and yet they are utterly unprepared to do so. And there is nothing that will happen in the next year or two years that is going to make them better prepared than the governments actually are, despite the talent gap.
So we can't abdicate the role that the governments have. Even as AI looks like an increasingly techno polar space as opposed to unipolar or multipolar. So again, I thank you very much for inviting me. I'm sorry I'm running to an airport right now, but I'm sure we'll have a chance to chat again.
12:03
12:03
Lord Alderdice
-
-
Lord Alderdice
Thank you, Lord chairman. Can I ask if within us tech and policy circles, there's any sense that by not leading on global regulation of AI, the US is ceding ground to some extent to China, which is making efforts in that space? In other words, can the US balance its expressed desire for technological dominance with an agenda to shape global rules and norms for AI?
12:04
Speaker 2
-
Speaker 2
Yes. This is something that we have been thinking about at CSIs for quite some time. We were thinking about this, you know, a year ago and recognising that the fact is that if the US shows up at a international engagement and does not have some sort of affirmative vision for AI governance, it's a very untenable situation because there is a strong desire for for other countries to figure out an approach to AI and that approach to governance.
And that approach can't consist simply of letting industry do its own thing that is not responsive to the political and popular pressures that that exist in a lot of those countries. And China is very cannily seizing on this. So if you look at how it's approaching its international and multilateral engagements, some of the rationale for this organization that set up Waco is explicitly around the fact that a lot of parts of the world, a lot of countries in the world, are not being included in discussions of AI governance.
To the extent that those discussions are happening, they're happening in the OECD or the G7. And so a lot of the a lot a lot of parts of the world are, are not being included. And there and China's making it explicit pitch here. And on top of that, they are taking a different approach to regulation. So they're you know, up to this point, they have largely regulated AI internally using existing legal frameworks, but they're now working on a comprehensive national framework.
And that is also part of their strategy. They're going to going to use that national framework when it when it's implemented in their international discussions and say that we recognize the seriousness of this issue, we recognize that governance governments need to be involved, and we have an affirmative vision for how that can happen. So that's our already happening. And then we have this additional dynamic, which is that we now have multiple repeated and ongoing instances of agents from US companies engaging in these autonomous breaches.
They're escaping the sandboxes that we're supposed to keep them isolated. There are instances where they have access in an unauthorized way for foreign government websites. And so now I think it becomes even harder for the US to go into an international engagement or forum and say, we actually think that the industry is capable of self-policing, that we should let them continue to develop with largely without restrictions, because now we are seeing incidents and and happening in the real world.
And while there haven't been significant damages that we know of to date, it's very easy to envision the pathway to getting from what is happening now to some sort of serious incident caused by an agent under development at a US lab. And so I think now the pressures from for AI regulation and AI governance are increasing all around the world. And that makes the the US lack of action on this increasingly difficult when it when it does international engagements.
Thank you
And that approach can't consist simply of letting industry do its own thing that is not responsive to the political and popular pressures that that exist in a lot of those countries. And China is very cannily seizing on this. So if you look at how it's approaching its international and multilateral engagements, some of the rationale for this organization that set up Waco is explicitly around the fact that a lot of parts of the world, a lot of countries in the world, are not being included in discussions of AI governance.
To the extent that those discussions are happening, they're happening in the OECD or the G7. And so a lot of the a lot a lot of parts of the world are, are not being included. And there and China's making it explicit pitch here. And on top of that, they are taking a different approach to regulation. So they're you know, up to this point, they have largely regulated AI internally using existing legal frameworks, but they're now working on a comprehensive national framework.
And that is also part of their strategy. They're going to going to use that national framework when it when it's implemented in their international discussions and say that we recognize the seriousness of this issue, we recognize that governance governments need to be involved, and we have an affirmative vision for how that can happen. So that's our already happening. And then we have this additional dynamic, which is that we now have multiple repeated and ongoing instances of agents from US companies engaging in these autonomous breaches.
They're escaping the sandboxes that we're supposed to keep them isolated. There are instances where they have access in an unauthorized way for foreign government websites. And so now I think it becomes even harder for the US to go into an international engagement or forum and say, we actually think that the industry is capable of self-policing, that we should let them continue to develop with largely without restrictions, because now we are seeing incidents and and happening in the real world.
And while there haven't been significant damages that we know of to date, it's very easy to envision the pathway to getting from what is happening now to some sort of serious incident caused by an agent under development at a US lab. And so I think now the pressures from for AI regulation and AI governance are increasing all around the world. And that makes the the US lack of action on this increasingly difficult when it when it does international engagements.
Thank you
12:07
12:07
Lord De Mauley
-
-
Lord De Mauley
Thank you, Lord chairman. AI governance, to the extent that it happens, would need to contemplate addressing a range of different strands of policy from, for example, the development of technical standards right up to arms control for autonomous weapons systems. Are there specific areas where you think, despite some of what's been said today, that the U.S. might seek to pursue a multilateral agenda?
12:08
Speaker 2
-
Speaker 2
Yeah. So I think there's some areas where it's very unlikely that the US will agree to some sort of multilateral approach, at least under the current administration. So that would involve, I think, any sort of broad treaty, arms control treaty, AI governance treaty, the creation of a new international governance organization to oversee AI. There are areas where I think the US is more likely to engage. I think the number one area there is around technical standards, so technical and other standards related to AI governance.
And so I think this is an area where not only has the US industry indicated a strong desire to engage, it aligns with the preference for this administration for things to be developed by a sort of company, led by a process that is led by technical experts from companies, and that is essentially driven by consensus, that includes stakeholders from industry. And so I think that is the most likely. Place where the where the US government will engage.
There are other areas that I think could be venues for engagement. I think we will continue to see the US sort of think about packed silicon, how to expand its. Its desire to to secure a domestic supply chain for a full supply chain for AI technology, starting with the critical minerals and going all the way to the final models. And related to that, I think there will be continued interest from the US in coming up with a export control regime that partners with allied countries around security issues, and works to prevent China from accessing not only the most powerful chips that underlie AI training, but also some of the semiconductor manufacturing equipment that is necessary for Chinese chip manufacturers to reach the same level of efficiency, performance and yield as we see with us designed chips like those chips made by Nvidia.
And so I think this is an area where not only has the US industry indicated a strong desire to engage, it aligns with the preference for this administration for things to be developed by a sort of company, led by a process that is led by technical experts from companies, and that is essentially driven by consensus, that includes stakeholders from industry. And so I think that is the most likely. Place where the where the US government will engage.
There are other areas that I think could be venues for engagement. I think we will continue to see the US sort of think about packed silicon, how to expand its. Its desire to to secure a domestic supply chain for a full supply chain for AI technology, starting with the critical minerals and going all the way to the final models. And related to that, I think there will be continued interest from the US in coming up with a export control regime that partners with allied countries around security issues, and works to prevent China from accessing not only the most powerful chips that underlie AI training, but also some of the semiconductor manufacturing equipment that is necessary for Chinese chip manufacturers to reach the same level of efficiency, performance and yield as we see with us designed chips like those chips made by Nvidia.
12:11
12:11
Lord Bruce of Bennachie
-
-
Lord Bruce of Bennachie
Thank you. Although the question I I'm required to ask is if the US is to drive a multilateral agenda on AI, which countries will it view as key partners? Is the UK well placed? But everything you said and Ian has said has made me question whether how much we want to be a partner, because Ian at the beginning said US must win and win. Win means winning everything. But equally the UN wants to win but having no control over it.
And then we're looking for partners and we're in a difficult situation being told you only partner with us on our terms and you mustn't deal with China, or should we be looking at ways to ensure that we have sufficient autonomy and innovation of our own and find a way of dealing with America as well, because it's, you know, the picture you paint is of a huge control or engagement, which is trying really to gobble everything up and exclude everybody else from the benefits.
So the real question is, how does the UK position itself with small, but we're not the smallest, and how do we do that in a way that still maintains some proper engagement with the United States and indeed helped, perhaps to get the United States to be multilateral about this and recognise the case.
And then we're looking for partners and we're in a difficult situation being told you only partner with us on our terms and you mustn't deal with China, or should we be looking at ways to ensure that we have sufficient autonomy and innovation of our own and find a way of dealing with America as well, because it's, you know, the picture you paint is of a huge control or engagement, which is trying really to gobble everything up and exclude everybody else from the benefits.
So the real question is, how does the UK position itself with small, but we're not the smallest, and how do we do that in a way that still maintains some proper engagement with the United States and indeed helped, perhaps to get the United States to be multilateral about this and recognise the case.
12:12
Speaker 2
-
Speaker 2
Yeah. So so overall, I think the UK is actually amongst the best positioned of countries in the world to engage on these issues in a way that that shapes what happens. I think there are some tactical things that are important for the UK to do, and then some strategic things that it can think about as well. So on the tactical side, I think these relate to things like we know, for example, that AI companies have these very powerful cyber models that they're releasing in a limited, they have essentially limited access or trusted partner ways of releasing those models.
And so these are models that are not made available to the public. They have very powerful cyber capabilities. Some of the cyber guardrails have are not in place on these models. And so they are very good at detecting vulnerabilities. So first , it will be important for the UK and other similarly situated countries to work with the US to be able to secure access to these models so that it can engage in testing of its cyber infrastructure and be able to remedy gaps in the in the approximately 4 to 6 month or or 4 to 9 month gap we see between when these models are when these capabilities are developed in US models, and then when they become broadly available through open weight Chinese models.
So that's one important thing. I think. Second is working with the US government and with US companies to figure out how to ensure that the AI solutions that the US government is able to procure meets its sovereignty requirements. This is things like ensuring that the UK government has control over its data, that it can ensure that the data, if necessary, won't leave certain controlled virtual or physical premises. Right.
Making sure that US companies, to the extent that the UK government wants to procure technology from them, are able to meet the sovereignty requirements necessary for the UK government to employ AI technology at scale. So those those are tactical things. Then we can get to the to the strategic picture here. I think the the thing that the UK does that has has been very forward looking is the establishment of the UK Security Institute.
It is the premier sort of institution run by a government for testing AI models. And I think it will continue to play an important role in advancing the frontier of of AI science, particularly AI science related to benchmarking and evaluating models. And I think that it could very well play a long term role in any sort of verification regime that we might see the US and China come up with.
So I can envision the UK Security Institute, for example, playing a very significant role in serving as an independent, trusted , third party arbiter of any sort of agreement between the US and China to slow down models or to. Or it can provide sort of results about the capabilities of models that are equally trusted by both countries. And so I would think about how the the UK can position its expertise in benchmarking and evaluating models to not only influence discussions around AI governance going forward, but actually be an implementing partner and part of the trust architecture that will be necessary if we see any, any agreement between the US and China.
And so these are models that are not made available to the public. They have very powerful cyber capabilities. Some of the cyber guardrails have are not in place on these models. And so they are very good at detecting vulnerabilities. So first , it will be important for the UK and other similarly situated countries to work with the US to be able to secure access to these models so that it can engage in testing of its cyber infrastructure and be able to remedy gaps in the in the approximately 4 to 6 month or or 4 to 9 month gap we see between when these models are when these capabilities are developed in US models, and then when they become broadly available through open weight Chinese models.
So that's one important thing. I think. Second is working with the US government and with US companies to figure out how to ensure that the AI solutions that the US government is able to procure meets its sovereignty requirements. This is things like ensuring that the UK government has control over its data, that it can ensure that the data, if necessary, won't leave certain controlled virtual or physical premises. Right.
Making sure that US companies, to the extent that the UK government wants to procure technology from them, are able to meet the sovereignty requirements necessary for the UK government to employ AI technology at scale. So those those are tactical things. Then we can get to the to the strategic picture here. I think the the thing that the UK does that has has been very forward looking is the establishment of the UK Security Institute.
It is the premier sort of institution run by a government for testing AI models. And I think it will continue to play an important role in advancing the frontier of of AI science, particularly AI science related to benchmarking and evaluating models. And I think that it could very well play a long term role in any sort of verification regime that we might see the US and China come up with.
So I can envision the UK Security Institute, for example, playing a very significant role in serving as an independent, trusted , third party arbiter of any sort of agreement between the US and China to slow down models or to. Or it can provide sort of results about the capabilities of models that are equally trusted by both countries. And so I would think about how the the UK can position its expertise in benchmarking and evaluating models to not only influence discussions around AI governance going forward, but actually be an implementing partner and part of the trust architecture that will be necessary if we see any, any agreement between the US and China.
12:16
12:16
12:16
Lord Robertson of Port Ellen
-
-
Lord Robertson of Port Ellen
I think the further questioning might be might be seen as being redundant. But anyway, we're very, very grateful to you for taking your time at this ridiculous hour in the in the morning for you, and we'll send you a copy of the transcript. And if there's any minor additions that you that you feel obliged to put, then will gratefully receive them. But in the meantime, thank you very much for your attendance.
12:17