Hi, I'm Paul Shorey, Director of the Technology and National Security Program here at the Center for our New American Security, with another edition of our Tech Program podcast series. I'm joined today by Cara Frederick, Associate Fellow in the Tech Program, and our very special guest today, Richard Fontaine, the CEO of the Center for the New American Security. Welcome, Cara and Richard. Thank you very much.
Thanks Paul. So we're here today to talk about your new article out in the Wall Street Journal, the Autocrats New Toolkit. I'll say catch you tight on it, will it ominous? Tell us what is the Autocrats New Toolkit?
Well, maybe I'll start off and then hand it to Cara to go into a few of the specifics here, but we wanted to look at technologies that are only now coming to fruition or that are not yet fully mature and look at how autocrats today, and more particularly tomorrow might use those for essentially illiberal ends to surveil their population, to meddle in democracies abroad, to perfect their own central planning and things like that. And so in this piece, we look at the rise of technologies like automated bots at scale, advances in natural language processing, artificial intelligence applications, facial recognition, and a number of other things that together can be used to help autocrats essentially do their jobs. And then with sort of as a spur, we believe to the democracies thinking about how to respond, but let me turn this over to Cara. Yeah, and I think it's particularly relevant today, especially with the growth of a lot of these AI-driven technologies, right?
They're going to help intensify scale and deliver things like propaganda with a lot more precision. So while autocrats have been using some of these ideas to oppress our populations and over centuries, I think we're sort of entering the next generation of what is going to sharpen their ability to do these things, and the answer is in a lot of these emerging technologies. So we really wanted to draw those out to highlight not necessarily how they're being used today, but how this could be utilized in the future and then at scale. I think that those are the most salient points of what we were trying to do in a circle.
That is, yeah, that's alarming. None of those sound like good things. Let's start with what is actually happening today. China is engaging in a massive campaign of oppression of Uighurs out in Xinjiang province.
There are a lot of methods that China is using that are very low-tech, but they are also incorporating some of these new technologies to assist in that campaign. Can you talk us a little bit about what the Chinese using the high-tech side to oppress its population? Yeah, so first I want to express a caveat here. We're not saying that China's already created this technological dystopia that is a panopticon of surveillance that everywhere you go, you're going to be alerted by robots and swarms are coming in things like that.
We do know that there is a mix of low-tech and high-tech measures to maintain control over populations, particularly in Xinjiang. And while you do have people embedded in certain households, keeping the human intelligence work to keep their eye on the people who live in those households and report back to authorities, you have that. But then you also have facial recognition technologies. They have stands at certain checkpoints where people who cross these checkpoints are automatically registered to a specific database.
And what we're concerned about is the fact that China really does seem to be posturing to gather and then potentially use all of its data to detect anomalies, detect patterns. Because what do these technologies provide other than tools to do so? We do think that given the history of human rights abuses by the Chinese government, that their goal for doing these things might be to single people out as potential dissidents and technology is giving them the capacity to do so. And just to amplify on that, so the Chinese clearly want to use large datasets first to collect it through the kinds of technologies we're describing, facial recognition, and their reports that they've collected DNA from the population out in Xinjiang and other ways of acquiring large datasets.
And then use AI-based applications in order to analyze it so that they can look for what they would describe as deviance. Now deviance is in the mind of the beholder when it comes to this. And so if deviance is described as departure from what the Communist Party of China believes, Han Chinese should be doing in their religious and political lives, then you're going to have a whole lot of deviance identified in Xinjiang. In fact, today you see a truly alarming number of Muslims in that region that are in essentially re-education camps.
Yeah, and it's not just confined to Xinjiang either. So I mean, think of what you can do now with when you have the botnets, right? But you can potentially automate. So you can do these things at scale in terms of micro-targeting.
You can do it with greater precision. So it's enhancing that precision, that scalability, the efficiency, and the breadth of some of these state-controlled content mills. That Harvard Research Report from a few years ago basically said that approximately 448 social media comments a year were turned out by some of these members of a so-called 10 cent or 50 cent army. So that's important.
And then I think what is going to be brought up is the social credit system. So China has explicitly stated that they want this to get this going by 2020. Right now we're not releasing indications that algorithmic scoring is in place. It's mostly reliant on these red lists and these black lists that prevent people from maybe getting high-speed train tickets or getting a government job.
But if you tend to think, OK, they have the system in place now, what is it going to look like in the future, especially as you get AI-driven technology like facial recognition to sort of bolster the surveillance capacity, then I think you can sort of make determinations as to where the state intends to go. I think that's what's remarkable about this is that you see that China is building actually the infrastructure built in facial recognition and things like social credit to build this very complex system of social governance. That China is currently using facial recognition, not just in China, but throughout the country in a variety of different cities to do things anywhere from catching jailwalkers to, in some cases, monitoring how much toilet paper people use in public restrooms. So that's even very trivial in fractions.
Well, and these kinds of tools are not just at issue inside China because they're increasingly available for export and they're not only available for export from China, but there are a number of companies around the world that are making the kinds of tools of surveillance or repression that certainly combined can have a massive difference on the way. Otherwise, liberal societies might conduct themselves. One example that has popped up recently is the sale of a facial recognition system and database to the government of Zimbabwe, which not only will it power that government in Harare to better surveil its population through this sort of mass facial recognition process, but the owners of the data will be in China. And so each of the Scandinavian faces will be exported back to China, which will then allow the company a question to refine its algorithm further and then have more opportunities to export it to different countries.
And so we're seeing whether it's Zimbabwe, certainly Russia, Venezuela, and of course China and other countries as well. We're seeing this as an increasingly pervasive phenomenon. And even in Gulf countries too, right? So what is the Chinese regime but a natural partner for countries that don't value free and open societies necessarily?
So what I'm concerned about is, yeah, is this going to start to look like a best practices exchange once the technical blueprints are transferred? So I think it's something to worry about the censorship and surveillance seminars that have occurred in some Gulf countries provided by Chinese experts. And these are not just Chinese government experts too. We do know that Chinese firms and companies are the ones who are selling a lot of this technology to other countries, but there is a degree of collaboration between the Chinese government and private companies that we don't really have here in America.
So when we think about what these firms and these companies are doing in other countries, we do have to think about the influence of the Chinese government on those efforts. I think that's what's so interesting about this. It's not just China, of course, building this at home, exporting the technology abroad, but as you secure, they're also exporting their laws, their policies, their model of how to use this because we have facial recognition technology being used in places in the United States. It's been used at Dulles Airport to catch people coming through airport security on fake passports.
But we have a different set of institutions in the US in terms of media, civil society, and independent judiciary that puts checks and balances on how this technology is used and transparency about where and how it's used. Well, this is key to the whole dilemma because these technologies, most of them are not in any way inherently bad, but like all other technologies, they can be used for multiple purposes. And the same facial recognition technology that can be used to spot a true criminal who would otherwise be violent, for example, in a crowd or for whatever other reason can be used to monitor and harass dissidents or for political reasons. And the same thing is true.
The other technologies we describe in this piece, whether it's automated bots at scale or advanced as a national language processing, Facebook is used natural language processing to try to get kind of an early warning when folks post something that might indicate they would be at risk of suicidal ideation. Well, that's arguably a good purpose, but if you're using it to keep track of people who are expressing political views that the regime in charge finds out that that's a very different use of the exact same technology. I totally agree. I think it's very important to recognize that there are legitimate economic benefits.
Like these technologies can raise people's standards of living and just make their lives better in general. So it's when they're repurposed for malicious uses. That's when we want to throw the flag. So what are these technologies?
The autocrites new toolkit. What does that mean for this broader one-one contest between freedom and authoritarianism? Well, this is a subject not just of this piece that we have written here for the Wall Street Journal, but a much broader project that the Center for New American Security is taking on that's looking at high-tech liberalism as we're turning it and will bring together efforts from across the organization to look at this contest and to determine what the democracies do about it. So that will bring in experts from our tech program and from our regional programs, from our economics program and so forth.
I think one of the important things to note here is that it's more than just a human rights question. It certainly is a human rights question. Americans have tended to care about internal repression in other countries and have wanted to try to do something about that if they can. But it's also about protecting our own democracies.
In 2016, we saw the cyber hacking of the presidential election in the United States. We also saw the rise of online bots on social media when Russian issues were being discussed here and there. If those become indistinguishable from human beings, then you have a different challenge on your hands. The 50-cent army that China employs is mostly desk workers that can post comments favorable to the regime or undermining of those who would oppose the regime.
When that process becomes automated and you're dealing with a different phenomenon, including a phenomenon that will reach the United States. Deep fakes on the video side are pretty immature now. It's hard to see really compelling content, but on the audio side it's right there. And you can immediately bring to mind a scenario in the future where a presidential candidate in the United States who might be critical of a foreign country has a video leak that is embarrassing or that somehow diminishes them.
And it will be potentially an open question about whether that video evidence is in fact accurate or not. And then of course the very fact that that's a possibility would allow someone to deny authentic video evidence. And so this is not just about improving democracy and human rights practices in other countries, it's about protecting the core democratic practice that makes the United States and other countries enjoy their own way of life. And that's one of the reasons why I'm so excited about this project too, because we're not just talking about how technology can be used to propagate a dystopian future.
We're talking about how you harness these technologies for good, right? So how do you do things maybe even defensively, right? So troll tracking, coding deep fake detection algorithms. So how do you use tech and turn that surge of liberalism on its face?
Well, it's very exciting. I'm also super jazz about this project that we've just launched here at C&S on countering high tech liberalism that brings together research programs from across the center. And I look forward to working with both of you on that. And let's just continue to hear more from us on that new project as it unfolds.
Thanks for coming to that.