Sen. Warner on the RESTRICT Act, AI, Bipartisanship on China, and a New Era of Intelligence

Also: Warner’s “classified roadshow briefings,” the “sledgehammer versus scalpel” divide in Congress, and a ChatGPT-powered IRS agency

Yesterday, I interviewed Senator Mark Warner for the podcast. It’s a rather extensive interview, so I’d recommend reading this transcript within the Substack app or in a browser.

Jordan Schneider: I’m honored to welcome Senator Mark Warner of Virginia, Chair of the Senate Intelligence Committee, to ChinaTalk. We’ll get into the RESTRICT Act, state capacity to analyze emerging technologies, the future of industrial policy, the nature and limits to bipartisanship around China, as well as the government’s role in regulating artificial intelligence.

The RESTRICT Act’s Origins

Jordan Schneider: So let’s start with the RESTRICT Act, your recent proposal to give Commerce the ability to get between Russian and Chinese ICT [information and communication technology] suppliers and US businesses and customers. What’s the path forward from a process perspective?

Mark Warner: Well, Jordan, before I answer that, I want to lay out my evolution on China. I remember when I was governor [of Virginia] back in the early 2000s leading a trade mission to China, very much encouraging more bilateral trade, encouraging greater academic exchanges — and then starting in about 2010, I got on the intelligence committee, where what I was hearing in the public domain versus what I was hearing in the classified domain were almost night and day.

This got much, much more contrast — the stark contrast when President Xi took over in the 2012 timeframe, where it became clearly evident that President Xi thought: if the CCP was going to maintain its leadership position, that he wanted 1) a more authoritarian regime, 2) that he saw this as a bilateral contest in terms of a lot of technology development, and 3) that he was going to basically ignore traditional rule of law and encourage intellectual property theft at an unprecedented level.

[This all grew] to the point that, starting in 2017, I started hosting a series of “classified roadshow briefings” — always on a bipartisan basis, with leaders from the FBI, different intelligence assets, the industry sector, [to talk] about this competition with China, as well as meeting with academic institutions.

And I want to make clear at the front end of this podcast — and I think it’s critically important to do so: my beef is with the CCP and Xi’s leadership. It is not with the Chinese people; it is not with the Chinese diaspora. I think when policymakers fail to do so, since so much of the diaspora communicates on platforms like WeChat, that becomes fodder for the CCP to say, “Hey, see? This is just anti-Asian, anti-Chinese rhetoric.”

Jordan Schneider: So, the bipartisan evolution and consensus on China is something you, as you just described, had a front-row seat to.

Were there particular turning points, and could you describe the nature of the consensus and how far it goes today when it comes to policymaking on the Hill?

Mark Warner: Well, I think as China changed its laws to make explicitly clear that all Chinese companies’ first obligation is to the CCP — not to shareholders, not to customers — that was important.

And the big wake-up call for me — and maybe as somebody who used to be in the wireless industry before I got involved in politics — the shot across the bow of Huawei coming in and not only being a leading player in 5G wireless development, but also China’s starting to move into the standard-setting bodies and literally flooding the zone on what had been normally standard-setting entities that (regardless of where the technology was invented) the United States dominated.

If anything, it became so bipartisan that — and particularly I would see sometimes on the Republican side almost a rush to see who could out-China-hawk each other — it has almost sometimes become problematic, because there are clearly places where we still need to engage with China; there are clearly things that we cannot solve on our own.

There’s got to be a level of sophistication. For example, I strongly don’t believe America should be buying Huawei telecom equipment because of the potential security risks of having the whole hardware stack running some of our wireless networks. But I’m still open to the idea of American semiconductor companies selling legacy chips to Huawei handsets, because that’s a commodity item that would otherwise be purchased from Korean, Taiwanese, Singaporean, or other entities.

So I do [try] to convince some of my Republican friends, “We’ve got to have a scalpel rather than a sledgehammer.” I cite the example [of] the Inflation Reduction Act: because candidly, there’s a series of areas — like solar, like battery power, and others — where we almost have to steal back or find a way to bring back technology from China if we are going to get the full advantage of some of those investments.

So I do think it is an area where there is broad bipartisan consensus. I spent some time with Mike Gallagher, who the Speaker of the House put in charge of this China Committee. I think Mike is a very smart and sophisticated guy. And [this broad consensus] is one of the reasons why I believe that the implementation of the CHIPS bill [being] done correctly is so important, because we’re into areas that, traditionally, some of my Republican friends were reluctant to go toward, which is quasi-industrial policy (where failure to have America invest or our allies invest will leave the field to Chinese entities, since they receive enormous amounts of government subsidies).

But if we don’t get the CHIPS implementation right, our ability to make similar kinds of investments — whether it’s an AI, quantum, synthetic biology, or advanced energy — is going to be seriously diminished.

So this is very much a work in progress, in terms of implementation around chips as we get into these questions about technologies that come from China, Russia, Iran, North Korea — it was the basis and the genesis of the RESTRICT Act.

The good news is: it’s broadly bipartisan. The bad news is that it’s a field that is changing in real time.

Jordan Schneider: Let’s stay on the CHIPS and Science Act. First: prospects for appropriations of the science part of that? — as well as what you think the CHIPS Act team needs to deliver on in order to execute in the way that will unlock future investments in other emerging technologies.

Mark Warner: Let me take those in order. I think on the science portion, there is a great deal of interest in the tech hubs — and the CHIPS and Science Act had a very aggressive twenty tech hubs; I’m not sure we’ll get to the full twenty, but I do think there are enough states around the country that are interested in potentially securing one that I see the prospects for funding there [are] pretty good. I’ve met with the CHIPS implementation team — I think they’re going about it in as smart and sophisticated a way as possible.

What I am immediately concerned with is, well, $52 billion sounds like a lot. ($39 billion of that will go as incentives, while $13 billion goes to research things like the defense industry.) The $39 billion, while it sounds like a lot — when you look at the size and cost of investments of these fabrication facilities or factories, I am seriously concerned that, [given] the big announcements that have already been made, the chip CEOs may come in and say, “Well, you know, I was thinking I only needed $2 billion, but now with inflation I need $3 billion” — and I’m gravely concerned that that money could all be eaten up just by the needs of the major announcements that have already been made. [Ed. Some of these announcements include TSMC in Arizona, Samsung in Texas, and Intel in Ohio.] And I think that would be a serious error. And I think the CHIPS team understands that. I hope they will make clear in their guidance, so the expectations are properly set.

We have to not only support the Intels and Microns and TSMCs and Samsungs, et cetera — but having this complete ecosystem, located in America and elsewhere but not concentrated in the PRC, means that that money has to be spent not just on fabs, but needs to be spent on other components in the supply chain (from the tools which are very important, to the packaging at the other end after the fad process).

And my fear is just, candidly, that the states that have already been winners will have senators and congressmen that are strong, strong advocates of their states, getting as much of this money as possible. And while the majority of states haven’t benefited, are enough of the rest of us going to be willing to say, “No, slow down — let’s make sure these investments are well made”? You’ve got to reserve for second- and third-round needs in terms of the supply chain.

Midjourney: corporate America’s naïveté vis-à-vis China

Corporate Naïveté and Avoiding WWIII

Jordan Schneider: Coming back to the questions around bipartisan consensus on what China is, where it’s going, and how the US should respond — you mentioned “sledgehammer versus scalpel” as one of the risks you saw of things you might be concerned about the GOP doing. I’m curious if there are any other trends? — as well as what potentially could happen within Beijing that might change some of your views on how the US should respond to China.

Mark Warner: Well, I do favor the administration continuing to talk. And I think things got beyond either side’s control in short[-term] sense — when Secretary Blinken was planning on going [to China], and then we had the incredible spy balloon fiasco. I think both sides wish those talks had been able to take place, number one.

Number two, I don’t see any evidence that Beijing, or President Xi in particular, has any indication that he is backing off. If anything, his statements in recent weeks have been more “pedal to the metal” about decoupling or concern about America rallying other non-authoritarian regimes into some kind of alliance.

I still get surprised at times — we have continued to do these “classified roadshows,” and at the most recent one about a month and a half ago, we had forty Fortune 200 CEOs in the room.

And one Fortune 50 CEO in this classified setting was basically saying, “Hey Senator, come on. You don’t really believe that Xi is going to move on Taiwan. You don’t really believe this is going to get to a crisis state.”

And I felt like, “Oh my gosh. You’re supposedly a sophisticated CEO; you’ve got a company with huge international holdings. If you are not building into your business calculus the chance that Xi could take on military intervention — or even short of that, a major blockade — and that the possibility of things ratcheting up is frankly, at this moment in time, higher (I believe) than things cooling off…” — it was a bit of a wake-up moment for me: there’s still a lot of people that don’t want to believe the facts.

I also saw this very much so pre-covid from the American private equity businesses who were making so much money investing in Chinese tech: they didn’t even want to take the brief — and they didn’t take the brief pre-covid. Post-covid, private equity has taken the brief, and I think they’re more understanding of the risks — but it’s still a shifting field.

Jordan Schneider: Do you have a vision for how we get to a steady state where you don’t have any of those scary scenarios, and the US and China are able to end up living in a world without World War III coming down the pipeline? Where can we end up that’s acceptable for both sides?

Mark Warner: The first place we need to start is to build up our alliances and not turn this into a binary choice where countries are positioned as, “You gotta take the American side, or you gotta take the Chinese side.”

There are a lot of openings now with a lot of nation-states. One, we’ve seen many of the countries that have been recipients of the Belt and Road projects realize, “Hey, this deal sounded too good to be true — and it was! It wasn’t that great a deal. We didn’t get the workforce because China would bring in their own workers. The quality of the infrastructure investments was pretty poor.” And the debt levels that a lot of these nation-states have picked up are terribly frightening. So with the Belt and Road countries, there’s an opening.

I think the way that Xi, through the covid lockdown, scared a lot of nation-states [provides an opening]. I think the treatment not only of the Uyghurs but of the people in Hong Kong in particular has had set off an alarm. And I think the fact that, throughout this period, the CCP has continued to have massive intellectual property theft — while they continue to do more and more locking down their own industries — gives us an opportunity.

And part of that opportunity is even how we talk about things. I cringe when any American policymaker says, “Well, in regard to China, America, and the West,” or, “America and NATO” — because every time we do that, we piss off two-thirds of the world. I was just in India. India is — appropriately, as having a large, long border with China — obsessed about China. [I think about] most of these nation-states in Asia — I think about the switchover even in the Philippines leadership in the last year, how their views on China dramatically changed. I think about all these nation-states in Africa, in South America, some of which were recipients of Belt and Road initiatives, that would much rather do business with us. But we’ve got to reach out.

So stage one is, “Let’s build as strong and broad an alliance as possible of non-authoritarian nations against the Chinas and Russias and all.”

I think where we obviously have to collaborate is around climate change. And China is trying to do both worlds at once: continue to invest in and move forward fossil fuels — but also at the same time invest at an even greater level than us on greener energy and advanced energy; I’m following what they’re doing in fusion energy and small modular nuclear. So I think we have to figure out where we need to collaborate; I think we need to keep these channels open.

I can assure you that one of the things that concerns me greatly (and again, the balloons was one example): China in certain ways is obviously a great, great nation — but the communication structure we have with them is still a little fragile. We’ve had seventy years of animosity with the Soviet Union and Russia: there are many touchpoints beyond even the direct Washington-Moscow hotlines where, if things start to go awry, there are communication modes to try to put things on pause. We saw during the balloon phenomenon: as I and (as has been reported in the public domain) Secretary Austin [were] trying to reach out to Chinese counterparts, [as well as] a whole host of individuals down the chain of command trying to reach out — and nobody in Beijing pick[ed] up the phone. That kind of immature response is one of the ways that could really lead to the kind of crisis where, [under better circumstances,] more reasonable heads could find ways to diffuse [it].

So, let’s build the alliance. Let’s find out areas where we need to communicate. Let’s make sure that we have open communication modes to different levels in the Chinese government, so we can avoid the unintentional crisis.

And then another area where I see almost as dangerous as any particular item is: as these new technologies develop, how do we make sure the standard-setting bodies, the policymaking entities that usually we’ve dominated — that China doesn’t dominate them? With all the potential upside and downside of AI — and AI that has a Chinese underpinning, with their lack of transparency, lack of respect for human rights, lack for respect (in what I think most of the rest of the world would agree) in basic human-dignity values — if that possibility doesn’t scare you, it should.

Jordan Schneider: Let’s stay for a second on this idea of a technological and economic equilibrium with China. You mentioned that you were fine with us selling the inputs to legacy nodes. Can you expand that more conceptually? What kind of framework do you think is okay for the US and China to be engaging in, from an economic and technological perspective, for the next decade?

Mark Warner: I think we have to make sure that traditional and non-traditional communication modes are available, number one.

Number two: I don’t see — from China’s activities to the statements of President Xi and other Chinese leaders — anything that indicates these technology competitions [are] based on a collaborative model.

Most of Xi’s statements — and some of the senior leadership — have been about Chinese dominance, about winning the competition. And winning the competition, in my mind, is not a signal that you send if you’re trying to say you actually want to develop together, or come to some common understanding, or recognize there ought to be this free flow of goods.

One of the things that has just amazed me for years — and I give again the CCP credit — [is] their ability to play off American business versus European versus Japanese, when they were stealing intellectual property from all three; [they were] able to play us off when we should have been alive against this unlevel playing ground. We should all be collaborating, but we haven’t. I see some of that changing, and I think a lot of this will require, again, the private sector to recognize [the CCP’s actions here].

What still bothers me is the number of large private-sector entities that say, “Yeah, I get that they are stealing from us, and I get the fact that they don’t abide by the rule of law; gosh, it’s awful what they do to some of their own people. But it’s such a big market. I gotta turn a blind eye.” I think that is changing — but [there needs to be] from business some level of recognition that this is, at the end of the line, not good for them if they’re participating in an economy that treats their people so badly. Or if the price of that joint venture is [that] you’ve got to give up intellectual property or you’ve got to transfer your technology — that’s still an evolving target.

Because we had both that alliance and collaboration with business — I think we then go much stronger into the vision of how the world’s two biggest economies can actually collaborate and work together.

How the US Can Lead in AI Regulation

Jordan Schneider: So this is a nice transition point to artificial intelligence. You and your staff recently wrote a letter to a number of leading AI companies and labs emphasizing this idea of research security and sort of asking them, “Do you have Kaspersky on your computers?” and a number of questions, mostly along the lines of the very reasonable question of, “What happens when weights get stolen via industrial espionage?”

It is interesting that OpenAI, when it started releasing products, had a number of countries already on the list — regardless of what export controls are now — including China that aren’t able to gain API access to its technologies.

I’m curious, what was the motivation behind that letter? — and more broadly, where you think government has a role to play in the regulation and potential support of artificial intelligence.

Mark Warner: Well, first of all, I’m trying to learn as much as I can, as quickly as I can — and the fact that I’ve got a technology background may give me a tiny advantage, but it is small — as I try to process all this myself, number one.

Number two, as you said: OpenAI and Sam Altman’s team — they are not participating with some of these nation-states, because China doesn’t want their people to be able to ask a question of a regenerative AI model, “Tell us about Tiananmen Square.” [The CCP] is all top-down, but don’t understate the amount of money and effort that China is spending trying to create their own platform here.

I’ve been educated that a number of computer scientists have said they’ve been able to hack into some of these platforms already and change data about themselves. One scientist talked about the fact that he made sure that anytime anybody queried him, they would put the word “cow” in there, and it’d say, “You know, Dr. X received all these numerous prizes, but unfortunately he has not done any major research on cows.” [Given] this ability to hack in and change the output, there should be at least a recognition that we need security built in.

For so many of our systems, we built [in security] after the fact. In these AI platforms, we need to build it in as the platforms are developing, not as an afterthought.

So I thought this was a safe, good spot to put a marker down, where I think there would be broad consensus.

In terms of a bigger role: I do believe there is a bigger role for government. I do believe there need to be some regulatory guardrails — the example of social media where we said, “Go forth, break things, and we’ll figure it out afterward” — we’ve not figured out anything afterward, and the US Congress’s performance has been pitiful in putting any guardrails in place. If that were the case in AI, we’re screwed.

But what that proposal [ie. future legislation regarding AI guardrails and standards] looks like, I want to be thoughtful, number one, and be better informed before I lay out some of those principles. And that’s the learning process I’m going through. And at least for today, on a good-news basis, we’re doing some of these “listen sessions” and “learning sessions” very much on a bipartisan basis.

Jordan Schneider: So this leads to questions about the executive and legislative branches’ ability to absorb, understand, and respond to emerging technologies. Over the past decade or so, you’ve had the US government get a lot more active on both sides of the promote and protect ledger.

Last year, you proposed the American Technology Leadership Act, which was an idea to create more analytical capacity within the executive branch to understand relative strengths and weaknesses in particular industries.

Bringing it back to your AI topic — we don’t have the Office of Technology Assessment anymore in Congress. Do you think there needs to be another layer of investment in analytical chops, both on the Hill and in the executive branch, to help policymakers wrap their heads around these fast-moving issues?

Mark Warner: The short answer is yes. I’d rather be the wingman on somebody else pushing that, because I think it’s necessary but not sufficient. And I do think we’ve seen, for example, in AI — a lot of the work is happening at OSTP [White House Office of Science and Technology].

And one of the things that annoys me, as a former VC and a former technology guy, is the disdain that most tech folks have for Washington — until they need Washington. Yes, we are an imperfect body to get all this stuff, but you got to help us, because if not, we’ll cede leadership — which we’ve done, for example, on all the things around social media from privacy laws to content moderation to kids’ online safety; we’ve basically turned over the keys to the Europeans or the Brits, who have advanced rules on all of those. So we can either lose our traditional leadership position — or, if you don’t work with us and we don’t do this in a thoughtful way, you get an overreaction or a simplistic approach that isn’t as nuanced as needed.

And sticking with the AI so far: many of the leaders that I’ve been talking with, I think, will be the first to acknowledge, “Yes, we need some guardrails. Yes, we need some government regulation.” The problem I’ve found, at least dealing with the big-platform companies and social-media companies — not in regard to AI, but in regard to Facebook and Google and Amazon, et cetera — is they’re normally in favor in theory, but when it comes to the written word, they’ve always got a problem. Even something as simple as data portability and interoperability — something that should be low-hanging fruit — we’ve not been able to get that done.

So I do think, back on AI, holding these folks accountable to actually work with us to get a thoughtful regulatory framework is what I’m shooting for. And the security piece is a place to start — I think there’s virtually 100 percent agreement. So how we translate that into a legislative package will be something I’ll be working on.

But, again, security is necessary but not sufficient — just as is the question about getting more expertise within federal government. Necessary but not sufficient.

Warner Defends the RESTRICT Act

Jordan Schneider: So let’s come back to the RESTRICT Act for a second. [What’s the] process forward? — as well as any changes you’re excited to make in response to the criticism you’ve gotten around the bill — in particular, with regard to open-source technology as well as the impact on individuals.

Mark Warner: Sure. Again, the genesis of this legislation was not about TikTok. It was about the fact that our approach to foreign-based technology that poses a national security risk has been, what I call, “whack-a-mole.” You mentioned Kaspersky — a few years ago, it was Kaspersky, and we’re still trying to deal with Kaspersky. More recently it was Huawei and ZTE. Now it is TikTok.

And what we thought was: we need a rules-based approach that recognizes — particularly when we’re talking about communication technologies there — we have got to protect the First Amendment that would give any of these entities their day in court; [and] it would be incumbent upon the government, including the intelligence community that I oversee, to declassify as much information as possible — so it’s not just, “The government says”; we got to prove our case.

And what was fascinating: this is not just a phenomenon that’s taking place in America. India banned TikTok three years ago at an individual-user level. Canada, the UK, the European Union have banned [TikTok from] government phones. More recently, the BBC told all their journalists to get off TikTok, because you could be potentially being monitored.

And again, the concern we have about TikTok is both the data collection possibilities — which ultimately might be able to be walled off, but so far have not. But equally, if not more important: [there are] 150 million Americans using TikTok on average, what TikTok says, ninety minutes a day. [We’re talking] about a misinformation, disinformation propaganda tool on steroids — as you see the numbers starting to say that about forty percent of people between eighteen and twenty-four get most of their news off TikTok [Ed. The statistic is forty percent of adults aged eighteen to twenty-four use TikTok, while fifteen percent use it to discuss or share news]. That, I believe, is a national-security risk.

Now, it doesn’t mean that we don’t also need to deal with Facebook and Google. I’m for a privacy bill, I’m for kids’ online safety and for Section 230 moderation. But this is uniquely different because of the CCP’s ultimate control.

And what we [Ed. see a list of Senate cosponsors for the RESTRICT Act here] crafted, we thought, was a very targeted act: [it has] no ability to touch an individual; [it] was not some massive expansion of government power. As a matter of fact, most of the language we took was taken directly from the rules surrounding the so-called “301 companies” that are put on watchlists already. So it was not something newly created.

And what was curious was to see: we were moving along merrily, picking up bipartisan supporters — up to twenty-six now, thirteen Democrats, thirteen Republicans; the Biden administration came out in support. And then the TikTok CEO, Mr. Chew, testified and had a (I think uniformly viewed as) pretty rough go at it.

And TikTok, which has spent over $100 million lobbying, flicked a switch, and suddenly there’s not a TV show you can turn on that doesn’t ultimately have a TikTok commercial on it; all the little political cheat-sheets that people on Capitol Hill read every day — the Politicos, the Axioses, et cetera — were all being sponsored by TikTok. We suddenly get attacks from the right and the left (some of the more liberal members in the House, and Tucker Carlson on the right) raising mostly completely false claims. But you spend a lot of money and you generate that buzz.

So we are very much looking at changes and talking to folks some of our critics to say, “How can we double make sure that, even if TikTok were ultimately banned, if an individual in American society wanted to use a VPN to get around it and get on TikTok, there would be no individual penalties?” — number one.

Number two, “How do we make absolutely sure that there’s not some massive expansion of government powers under a Patriarch 2.0?”

Number three, “How can we make absolutely clear that an American company that happens to be doing business or touching China in some way couldn’t be covered by this law?”

We don’t think any one of those three is truly justified — but part of the legislative process is to put additional belt-and-suspenders or make the changes to rebut these arguments.

At the end of the day, it’ll be interesting to see whether that will bring people along. Because one of the things I’ve said: Listen, there’s a lot of incredibly creative things going on in TikTok. There are a lot of people that now make their living as social influencers. I’m all for that. I do hope, though, and believe that if TikTok were to go away, [there would be another company] — and it doesn’t have to be an American company, there could be a Brazilian or an Indian or a Canadian company — that would provide those same earning opportunities and that same kind of creativity.

China and the Intel Community

Jordan Schneider: Let’s talk about China and the intelligence community. There’s been a lot of reporting about frustrations that Biden administration officials have expressed about their inability to get real, concrete answers from the IC — as well as some news around a new China Mission Center in the CIA, and a broader focus on transferring out of South Asia more toward engaging with questions around China.

I don’t know if you want to give a grade of how this has been going, or maybe some reflections on the challenges and what, if anything, needs to be done from a legislative perspective to keep that ball rolling.

Mark Warner: Well actually (and it’s taken longer than it should), we had our whole IC — that was generally focused on the terrorism threat for a decade — go back to great-power competition with Russia and China. And there were always at least the remnants of an operation vis-à-vis Russia, formally the Soviet Union. So it was almost easier to go back toward Russia. To focus on China has been a shift.

I can tell you the budget proposals this year are finally hitting the numbers that I would like to have seen last year. I am surprised there is this new China center [and] this new technology center at the CIA. I’ll be seeing intel leaders literally today — so I think they’re doing a pretty good job. But if they’re not, I’m making clear to the administration — and for that matter to Congress and the public — what those centers are focused on, and I’ll see if we can improve.

But the bigger question here — and it gets to a really hard policy issue: traditionally, the intel community has spied on an opposing nation’s military capability and political leadership. The problem in 2023, at least in my mind, is national security is a lot different in 2023 than it was in 2003 or 1993. So, simply who has the most tanks and guns and ships isn’t necessarily who the winner is going to be. And we’ve never faced an economic challenger like China, that is making the kind of investments in these technology domains.

I believe absolutely [that] national security includes who wins the race for AI, who wins the race to where we produce semiconductors, who has the overhead capabilities in terms of satellites, who does synthetic biology, who does advanced energy. And the nature of our spying operations: we always look out, but don’t look in.

So if I go to the ODNI — the [Office of the] Director of National Intelligence — and talk to their analyst: first of all, getting them to focus on all these technologies (rather than just the political leadership and the military) is a work in progress, number one. But even if they are focusing on China’s investments, and then we say, “And how are we doing in comparison,” they say, “We can’t look there.”

A Goldman Sachs first-year associate would know more about where we are in many of these technology areas than some of our best intel analysts.

Now, people are recognizing this. General [Paul Miki] Nakasone, who is, I think, one of the real leaders at the NSA that does a lot of our signals intelligence, has said he needs about thirty folks from Commerce Department to work on this. Secretary [Gina] Raimondo, I think, absolutely gets the fact — just as the CHIPS implementation is part of national security — that following these other technology domains is part of her job.

But how we marry that information in a way that doesn’t freak out the American people — “Oh my gosh, we’re looking at how America’s developing in these various technology areas” — is something we still got to figure out and then make the case that this is critically important to the American public.

Jordan Schneider: And is that a new body? New authorities? Is that an open-source center? Do you have a vision of what the solution is?

Mark Warner: I would say that’s also a work in progress. A lot of it will be around open source. We go back to the power of AI: almost all of that has come from open-source data. And, sometimes, convincing the intel community that it’s just as valid if it comes from open source as opposed to stealing a secret is still a little bit of a mind shift. But laying out that in a thoughtful way to 1) make the case most broadly that national security is much broader than it used to be, and 2) for us to compete — we need to really be looking at particularly China, where they’re investing and where they’re making their progress.

We have to also look at, then, how we and our friends are doing in comparison. And that’s a different mission from a cloak-and-dagger spy or a James Bond movie, focusing on a particular political leadership or who’s got the most advanced piece of military hardware.

Jordan Schneider: If you go back to some reports that came out in the 1980s between the US and Japan, you had the Office of Technology Assessment really doing great work looking industry by industry about where the US was relative to China and where the US could invest or not. Maybe we’ll get that again.

One last question for you: have you thought much about AI’s ability to improve domestic governance more broadly? How long should voters have to wait for IRS-GPT, and what is an appropriate lag between all of this stuff — improving productivity in the private sector versus that ending up impacting everything from intelligence collection all the way to more prosaic government services?

Mark Warner: Great question. As a matter of fact, Ian Bremmer, who follows a lot of international affairs and stuff, had a good piece today on this very question.

I have not seen somebody articulate that vision yet. I hope I can get informed enough to at least help or put out some ideas about this. But to be honest with you, if we were having this conversation a year ago, I thought at that point AI was simply an overhyped term for advanced statistics. Then I started meeting some of these groundbreaking leaders and going, “Holy heck — this is a lot.” And then I would even hear from some of the folks who’ve been the technology doomsayers and got to the point of being a bit terrified.

And [I’m] trying to now adjust to where I fit on that continuum, get informed in real time, and hopefully counsel my colleagues not to get so far in front of their skis on policy prescriptions that will either go too far or be laughed at as being naïve — [that’s] where we’re trying to hit a sweet spot.

And this question of the upside in terms of how citizens interact with government — the potential is great, but the potential for mischief is great as well.

Thank you so much for having me. Hope we can do it again.

Midjourney: ChatGPT meets the IRS