AI, Big Tech & Global Power: Oxford University Dr. Jennifer Cassidy on Diplomacy
Dr. Jennifer Cassidy of the University of Oxford, an expert on AI and diplomacy, discusses how artificial intelligence and Silicon Valley are changing the way states communicate, represent themselves and negotiate. She argues that big tech firms have become digital sovereigns that hold both hard and soft power, and considers former politicians moving into tech jobs, authoritarian uses of AI and the lack of binding global rules.
Transcript
At Donald Trump's inauguration, the most powerful figures in Silicon Valley stood just steps from the president. A clear reminder that technology now sits at the center of global power. Today, that same dynamic plays out all across the world. Former prime ministers hold senior roles in American tech giants. AI firms shape not just markets, but geopolitics. And the biggest players now wield influence rivaling nation states.
In recent episodes of Agents of Tech, we've explored how the UK's innovation ecosystem, from Oxford's AI research to Larry Ellison's 10 billion pound bet on tech and medicine, is positioning itself in this new global order. Now, as big tech takes on diplomatic roles once reserved for governments, the question shifts. Not who builds the technology, but who governs it? And can traditional diplomacy keep up?
This is Agents of Tech. I'm Autria Godfrey in Washington DC and joined by my co-hosts Stephen Horn and Laila Rizvi joining me from London today. Hey guys, pleasure to see you both. Well, this is a very exciting week because we are joined by Professor Jennifer Cassidy from the University of Oxford. She's an expert on how AI and Silicon Valley are reshaping diplomacy. You know, diplomacy used to be all about treaties and territory. Now it seems like it's more about data, algorithms and the companies that control them. Do you guys have any concern about politicians crossing over into technology? What are your thoughts?
I think there's just a complete change in kind of the world order, which we probably haven't seen in recent history, just because of the concentration of power in US private companies. And I think the nature of AI is that because it's a technology that grows exponentially, and at the a fast rate than any other technology has grown, that this is happening at a pace that we can't even imagine. So you feel like you're being left behind a little bit with it.
So I don't mind so much the people. I mean, this has always been the case, as far as long as I can remember, where your senior politicians go and get paid an awful lot of money by big companies to sit in their advisory boards and, you know, and to give status to, to those companies and, and get very well paid for that. I, I don't mind so much about the people. I don't really think that has changed hugely. I guess it is just the power, isn't it, of these, uh, IT, AI, call them what you will, companies, and how it pervades everyday life that just, you know, that just bothers me. And also the phrase foreign policy of Silicon Valley. What's Silicon Valley, a country now? Silicon Valley an empire? Why does Silicon Valley need a foreign policy? Do you have a foreign policy, or I...
I don't, I don't, and I don't think that our company does. But you're not wrong. I mean, I think Silicon Valley would love to hear, you know, that they're have been elevated to that level. But it, it does, it does beg the question, you know, when you're using AI with regard to diplomatic efforts, are, are we only going to see the, the divide between the haves and the have-nots widen, really, at an exponential rate? Who's really in charge of global power? Is it still those that we have elected to office, or is it these people coming in that are just being appointed to these positions?
Yeah. Right. And I think that, I think that's such a great question you raised there, Autria. And I think the answer is we don't know, right? I mean, we don't know who wields global power. We don't know whether it's Meta in your Facebook profile or whether it's the government. We don't know. And I think we're finding that out all the time.
I think it'll be a very interesting next 10 to 15 years to see how AI really reshapes, uh, the global world, really. But you know who does hopefully have the answer, and that is Professor Jennifer Cassidy. So, let's get right to it. Here's our conversation with her today.
Dr. Cassidy, thank you so much for your time today. We really appreciate it. I'd like to get started with kind of a scene setter of a question, big picture. How is AI reshaping diplomacy? Everything from negotiation to influence operations. And can you kind of lay out who you think the big key players are right now?
If you think of diplomacy, as I always do, as resting on three timeless pillars, that is communication, representation and negotiation, then what artificial intelligence is doing here is not demolishing these pillars, but quietly rewiring the architecture that holds them together. Now, something really important that needs to be said, and it's kind of the thesis that runs through, you know, all, all my work, is that every tech age has done this. So the telegram compressed distance, the aerial plane compressed geography, and Twitter, or X as we're calling it, um, or everyone's calling it, compressed time. Everyone makes these, you know, declaratory statements that this is the death of diplomacy. But, you know, from all research and all, you know, analysis, it is truly an evolution and not a revolution.
And AI is really transforming how states understand and respond to information flows. So predictive analysis now allows ministries to read the global mood almost, almost. They're, they're not quite there yet, I have to say, in real time, but we're certainly going to get there. And the examples here being, we have the United Nations and the World Bank. They're using AI models that monitor food prices, uh, rainfall patterns and social media data, data to anticipate instability in a number of regions that they're watching, really up to six weeks before that instability might actually break out. Um, we also see NATO employing machine learning, uh, to map Russian disinformation and detect coordinated inauthentic behavior long before it even gets to the diplomat's desk. So really what we're seeing here is the move from reactive diplomacy, which has always been, um, in place, to anticip, anticipatory diplomacy. So really the ability to foresee and preempt rather than simply response.
You obviously mentioned how we've, um, had technological revolutions all the time, and diplomacy has adapted to it. And there is a lot of hype around AI, obviously, that it will, you know, take over the world and it's like an alien species. But where, what are the areas within diplomacy that this time round it, it really is different?
So I would argue that we're still at the evolutionary phase. I could be saying, you know, in five years this is a revolution, but one area really is that is equally, um, or edging on quite, um, revolutionary is in representation and indeed negotiation. So in this sense AI can draft speeches, policy briefs and communiqués in seconds. But what's different here is we know in 2025 that no large language model is neutral. So what diplomats and foreign ministries have really trying to be, uh, scrambling with, just really in the last two or three years, is to make sure the large language models and the code that they're training their foreign ministries in is trained within their distinct political and econom, economic ecosystem. Really making sure that their large language models or their open-source, um, AI models are embedded with their own history, laws, and linguistical biases.
And that's why governments are beginning to, thankfully, I've been calling for this for like 10 years to make sure this happened, but to build their own, what I like to call, sovereign diplomatic AI systems. So what, one of the most pressing questions in this really re, revolutionary aspect of AI for diplomacy today is not how diplomats are using AI, but whose AI are they using.
So, so, so part of the problem with 21st century communications is we're in our own echo chambers, and we talk to people who talk the same way as, as we do. And isn't there a problem in what you've just said, if, you know, countries are following their own large language models, their own way of working, that they're just not going to pick up on the signals that are coming from other countries?
So what I was talking about there regarding the pillars of representation and negotiation, it is imperative that these nation states or regional blocks do have their own large language models or AI systems that reflect their cultural values, the, their historical lineage, that they can produce information. But this is not in any way stopping them to have the dual use of making sure that they listen and respond to what other nation states are producing. But just regarding the production of policy documents, communiqués, that still needs to be in the national or regional foreign policy voice of that state. But as you rightly noted, if we in, if we just decide to shut out all these other nations who are using it, we are removing ourselves from one of the great, um, attributes, um, um, positive, um, tools that AI is bringing.
So Jennifer, I wanted to ask you about power dynamics, right? Um, where we're obviously talking about the utility of AI and diplomacy, but does diplomacy kind of go out of the window when power, and in the form of very powerful AI models, is concentrated in private companies? I mean, we say, uh, France is training its AI, or the European Union, on Mistral, but Mistral is a French company. And the same with the US, as in the US's AI policy and AI models aren't really the US government's. They're OpenAI's AI models, they're Anthropic's AI models. So what does that mean for diplomacy?
We know that diplomacy has always been revolved around states. They, they've held territories, armies, and embassies. But as you rightly noted, in the 21st century the real infrastructure of influence has gone digital and actually now sits largely in private hands, most of them now at being in Silicon Valley. And if we think about it, um, the scale, like Google reaches more people daily than the United Nations, and its translation algorithms shape how billions understand the world and understand each other. And Microsoft's Azure cloud now hosts data for NATO itself, for multi-EU institutions and even for national governments, which is, you know, somewhat terrifying. So therefore these companies really have become what I like to call digital sovereigns, that is, private actors with public power, and they control the three levels of power, or levers of power, that were once, you know, purely defined by the state. That's information, infrastructure and interpreta, and, uh, interpretation.
How much of that power is, is soft power, and how much is, is it hard power? I mean, it sounds like quite a, you know, a strange concept, the foreign policy of Silicon Valley, just, you know, because these are private companies. I mean, WebsEdge doesn't have a foreign policy, you know. I'm sure University of Oxford doesn't have a foreign policy. So how much of this is based on the fact that the, just the economic power that they wield?
It's not that they have a distinct foreign policy from the United States, but they are, as we see with, um, you know, uh, defense companies, uh, engaging with, you know, autonomous weapons, all under the realm and all being contracted by the US government. So the blending of that, so it's not a distinct foreign policy in, in itself. Um, but all, but what you said is, is also spot on. Look, the world, we already mentioned the East India Company, we had, you know, the, um, uh, the oil industry, which obviously we still have, but after that we had, you know, the media mogul. These, these all gained huge amounts of power in, in their time, and this was purely as well due to the economic power they hold. What's different about, um, the Silicon Valley compared to the oil, uh, the, uh, media, is that they, these Silicon Valley actors hold both hard power and soft power. This is something we haven't seen regarding private actors entering the world, the diplomatic world, before. So it's a, it's a bit of both.
So, Jennifer, to bring things back here to Washington DC, you know, you talked about all of the, the tech oligarchs that were standing behind, um, President Trump during his inauguration, which is, you know, just steps from where I'm sitting right now. I would, I kind of want to flip that on the other side, though, and get your thoughts on some former political leaders moving into some of those senior big tech roles. So, I mean, you know, Nick Clegg at Meta or Rishi Sunak at Microsoft. Does that also blur the line to a, uh, concerning level between public service and the private sector?
It's more than a blur, I would say. I know I keep using the word quiet, but it's only quiet in the sense of it's not a public narrative we're seeing. Yes, for, for those of us following the story of AI and governments, this is not quiet. What used to be a revolving door between government and industry now just has become, I would argue, a circuit of influence. And this is where experience, access and authority really are just flowing continuously between capitals and campuses in Silicon Valley.
So you mentioned Nick Clegg. Um, so his move from Britain's deputy PM to, he's now Meta's president of global affairs. This is not, you know, any small role. And, you know, I would argue that this was not a just a career change. It was a signal. So he now performs many of the functions of a diplomat. If you look at, or if we, if we go back to what I was mentioning, that there are three pillars of diplomacy, communication, representation and negotiation, he performs all of these. So he represents an entity with billions of global citizens. He negotiates with governments, and he shapes the global conversation on regulation and, and freedom of, um, of expression. In many instances, we saw Rishi Sunak as well joining Microsoft as a senior adviser, and this sent a similar message as well.
On your point about, obviously, these lines being blurred, and I guess nation states becoming almost second tier when it comes to diplomatic conversations. But what does that world look like where private companies like Microsoft, like Google, are now those primary, number one tier, um, organizations or groups in those diplomatic conversations? What, what does that mean for society? What does that mean for us? And is it a bad thing, or are there some good things to come from that?
Uh, well, I don't think we're, it's like a PhD topic, but, um, God help the person who has to do the PhD on that. But look, look, it's, it's tempting to see this as business as usual, you know, just, this is how the world order is working, because, look, after all, we don't want to overplay this. Like, senior figures have always moved from public service and private enterprise. But what's different today is not the movement itself, but, you know, as you, as you alluded to, the magnitude of power that's being transferred when that movement occurred. So if we look at when, in a diplomatic perspective, when a retired general, you know, joins an army manufacturer, we know that influence remains within a defined sector. We know this. But when a former head of government steps into a tech company, they are entering an ecosystem that touches every sector at once. The economy, education, national security, even culture, and, you know, even freedom of speech, or what we see or what we don't see when we pick up our phones. So this is because we know that tech isn't a single domain. You know, it is now the operating system of modern life.
But, you know, I, I will say that there still needs to be nuance in this. This is not all bad. We need to remember that democracies do need experienced voices helping to steer the, the tech transition. We, we know this. Um, but we know that also experience is not neutral, and it comes wrapped in, um, networks of loyalty and language. So really, finally, the task for modern governance is not to ban the exchange, I don't think that's healthy at all, between public and private spheres, but to civilize this, really to ensure that when the expertise moves, accountability moves with it.
Can I just unpick that just a couple of ways? So one of the things that, so I'm interested in two things there. One is, you know, we talked about very much this from a western lens. You know, we're talking about Rishi moving to Microsoft, and we're talking about the, you, Silicon Valley. We're talking about the foreign policy of Silicon Valley. Is this exactly the same with Alibaba in, uh, in, you know, in China? Is it exactly the same with the big Japanese companies, um, kind of in, in Japan?
This is the same, and I'm just mentioning these, Nick Clegg and, and Rishi Sunak, as, as the examples, but it is no different when we're looking at other regional blocks or other nation states outside of the west. Uh, there is no UN or, or Geneva code, let's say, on a conflict of interests.
Right.
This is the revol, this aspect of revol, the revolving door between private power and public interest in the tech sector, it, it, it is very reflective no matter what, uh, regional block or what nation state you are looking at.
So, so arguably it was always said that, you know, post Second World War American power was, was spread abroad through, you know, the use of selling power, selling Coca-Cola, you know, and, and to, so to, so I'm interested in, in to what extent that's the same here. You know, ChatGPT is now accessible to UK citizens, who therefore, you know, become part of that US system, or to what point, you know, we're actually getting those companies exercising that power on their own.
At, at this moment in time, I don't think we're at a place to make, you know, a strong core thesis on, this is act, exactly what's happening.
Right.
Um, but what's different here in the tech aspect is perhaps the defining sovereignty dilemma of the 21st century is the private ownership of these tech companies, and that is not who controls the territory, but now who controls the infrastructure of decision-making itself. So if you think about it, really the core machinery of modern power, that is data, computation and communication, this no longer belongs solely to governments, but increasingly sits in the hands of the, of private firms whose platforms have become as strategic as a, as any weapon system. And you might think, oh, still, isn't this akin to these private companies of Coca-Cola or Hollywood and their sphere? No. The extent of power, and as I mentioned previously, the fact that these tech companies and their, their products, their actions, their policies are, you know, planetaria. They rep, they are conducted in boardrooms but have the, have implications worldwide. This is highly distinct from what we're seeing in the 80s and 90s with these private companies.
So Dr. Cassidy, not to play the what if game, but, you know, these tech companies have really been advocating how AI is going to be used for good. Um, you know, helping with food insecurity around the globe, uh, combating the climate crisis. But could AI, with regards to diplomacy, perhaps enable new forms of authoritarian control, or something we've discussed a lot here at length in the United States, election interference? Where does your level of concern lie with AI being used for diplomatic efforts, but by bad actors?
Before I just, you know, get into some of the examples here, it's really important to note, and it may seem like this kind of most basic statement, but how AI is used is all down to the intent behind the, the intent of the actor that is, that is using it. So we, I, when I'm giving these examples, I don't want to say that it is AI doing this. It is the intent of the actor who is using it. So, you know, to, to jump in there, we're seeing that, look, regarding authoritarianism use for, um, AI, or is it, or, or democratic, more democratic use? We see both outcomes that are already happening side by side. You know, in, in, in China, for example, predictive policing algorithms are tracking not just where crime might occur, but who might commit it. Again, this is anticipatory diplomacy, which we were speaking about, um, earlier, but in a, you know, under, under the, the, the framework of authoritarian regimes. And they're also combining facial recognition, travel data, and digital behavior into vast surveillance scores, which is, you know, terrifying to say the least.
Um, and we also saw that, you know, during the 2024 Iranian protests, state authorities used AI-driven sentiment analysis to monitor online speech and preempt demonstrations before they even begin. And as you rightly noted about election interference, you know, we've already seen this being proved and, and play out in elections, uh, both in the US and also around the world. So this is digital authoritarianism in its most refined form. It is control by prediction rather than force.
So in democracies, well, you know, in a, in a thriving democracy, we see many states calling themselves democracy, this my personal opinion at this moment, but, you know, in a thriving democracy, by, if we're looking at the structure of Robert Dahl, the scholar, you know, a, a democracy would ask regarding the use of AI, well, what do my citizens need? How can I use this to help them? But in, for authoritarian regimes, they would ask, well, what are my citizens going to do in order to, you know, perhaps destabilize the regime, and how can I use predictive analysis or facial recognition in order to stop this before it even begins?
Oh, that was going to be a quick followup for you. Is, given that we know the potential is there for AI to be used in this type of, you know, weaponization manner, do you feel like the, the big tech companies are doing enough to put the guardrails in place to try to counteract that? Or was it kind of a, hey, we put the, we put the technology out there, what happens happens?
No, not at all. This is the truly the, the weakest, um, and I would say darkest point. It's not very, not very, um, positive, and the biggest hill to climb when we're looking at AI diplom and diplomacy, with every single thing we've discussed so far. If we had a true governance and ethical framework which was global for the use of AI, a lot of these answers, and a lot of, uh, to these questions may look a lot different. The, the lack of binding framework that we have for the use of AI at, um, you know, a governance level is pretty terrifying. And, and the only thing that we have on a binding level is the EU. It's a regional one, the EU AI Act, which just came in in 2025.
Now I will just say on to, on a positive note at the end, that, you know, the international community is trying to do this. You know, we have the UN advisory board for AI. We had the, uh, G7, uh, Hiroshima process for AI, but none of this is binding. So it's simply just countries signing up and saying we will adhere to these structures better. So we have a very, very, very long way to go regarding the governance and structure, um, of, and frameworks for AI. But I will say that a lot of people say, oh, this, this is going to take too long, because we know the speed of technology, and when we look at the speed of AI in particular, how on earth is, is regulation going to keep up with this, the, the speed of that, that transformation? But I would very much press the point that is, yes, it's going to be a difficult task. It is a difficult task, but it's one that has to be done.
Absolutely. And the fact that we're even discussing it now, thank goodness.
Yes. Yes.
Well, I think it's good that we finished on a positive note. And thank you so much, Jennifer, for your time.
No, thank you for having me.
Okay. Well, at least we were able to end on a positive there. Uh, because, uh, Dr. Cassidy certainly, I think, speaks for a lot of, a lot of people, a lot of, uh, scholars in particular, who have a big concern about the way that, that AI, uh, is going to be utilized potentially by some of the, the bad actors out there. What stood out to you both about our conversation?
I think for me it was, um, her, what, what she was kind of talking about at the end, which is the, the pathway to potentially an authoritarian kind of regime, or totalitarian regime, which, um, in terms of how various states are using AI for good and for bad. Um, but I also think it's interesting on the flip side that people also argue that governments who are trying to aggressively enforce AI regulation are using kind of that, um, control to restrict people's freedom to innovate. So I think you can take it on both sides in that respect. So I thought that was pretty interesting.
I think it's just raises so many questions, you know, and, and, and, and it seems to me that you can label kind of, uh, actions from different people in the, in the kind of way that you want. So, you know, that, uh, she was talking about in China, how the, the, the police will use, uh, uh, AI to predict, uh, you know, behavior, and, and, and, and then arrest people, or, or whatever, according to that behavior. Well, it might not be quite the same in some western countries, but not a long way off it. And so, how you draw these particular lines is going to get very, very interesting. As you say, Autria, not, I wouldn't give it 10 to 15 years. I'd say next one to five.
You're moving at the rate of AI over there.
You know, us.
Yeah. Stephen, I, I thought your question about, you know, some of these countries building their own large language models to operate solely on the data of their particular culture, really creating this echo chamber where they're, they're trying to advance really only their interests. I thought that was a really, really good question to ask, and her, her answer to it was, was kind of interesting. She was kind of like, "Yep, that's kind of, uh, something we have to be cognizant of."
I think, is it Albania that has their AI minister, right? Have you seen that, the, their virtual digital AI minister? I mean, it's crazy. Um...
But yeah, so there's no reason why they, they wouldn't train on their data, their culture, and then start putting out these AIs which are AI diplomats. I mean, we didn't even get to that, that point with...
AI. Oh, it's a bot. Oh, how interesting.
Let's get, bring Jennifer back. Put that to her.
So much more to discuss still. Well, a huge thank you to Professor Jennifer Cassidy for joining us today. By the way, if you've got thoughts on how Silicon Valley is shaping global politics, whether you believe they are or not, or whether big tech has outgrown traditional diplomacy, be sure to drop them in the comments. We would love to hear from you. Weigh in. And while you're there, make sure to hit subscribe and like, and we will see you next time right here on Agents of Tech. Stephen, Laila, always a pleasure to see you. Until next time.