The “Don’t Cry for Me, Lawfare” Edition

Summary

Rational Security – “Don’t Cry for Me, Lawfare” Edition

Date: January 29, 2025
Hosts: Scott R. Anderson, Quinta Jurecic, Alan Rozenshtein
Guests: Kevin Frazier, Eugenia Lostry, Peter Harrell

---

Episode Overview

This episode of Rational Security centers on three major national security and foreign policy developments:

1. Trump administration’s escalating international spat with Colombia over migrant repatriation flights, its resort to sanctions and tariff threats, and what this “America First” moment signals for U.S. diplomacy and global partnerships.
2. The United States’ high-stakes bets on artificial intelligence, including Project Stargate, and China’s disruptive advances (specifically Deepseek’s open-source AI breakthrough), with discussion of how these shape global technology competition.
3. A sweeping Trump-ordered freeze on U.S. foreign assistance, including its deep impact on development, security, and international standing, especially as China steps in with an alternate model.

Noting that this is Eugenia Lostry’s farewell episode, the discussion is both reflective and rich with expert insight, peppered with characteristic Lawfare wit and memorable moments.

---

1. "Tariff or Takeoff": U.S.-Colombia Diplomatic Incident

[05:01–33:47]

Key Points

• The Spat: The Trump administration attempted to deport migrants to Colombia on military flights. Colombia’s president, Gustavo Petro, refused the flights, accusing the U.S. of disrespect and even racism. Trump’s team, caught off guard, fired back with threats: tariffs, visa restrictions, sanctions.
• Swift Back-and-Forth: Petro reversed course within hours after intense pressure, even retweeting the U.S. summary of the call—each side claimed victory, with mutually incompatible narratives.
• What Was at Stake: For Trump, military repatriation of migrants is a political priority, and symbolic pushback by foreign leaders can’t be tolerated—domestic optics were important. For Colombia (and similar countries), dignity and treatment of deportees were central concerns, as was signaling some independence from Washington.
• Diplomatic Style: The episode highlighted the return of foreign policy by social media and the unpredictability—or calculated brinkmanship—of the Trump approach.

Insights and Analysis

• Peter Harrell: "We are definitely back to the era of foreign policy by social media post." ([08:53])
  • U.S. staff were surprisingly unprepared; typically such flights would be negotiated well in advance, avoiding these confrontations.
  • Short-term "wins" (i.e., Colombia backing down) can come at the expense of long-term relationships: “If you tick these guys off too much, well, they might cave to you in the short run, but over the mid and longer run, your partners may drift away from you.” ([14:36])
• Eugenia Lostry: Emphasized that Colombia’s main objection was dignity, and some concessions were actually achieved: “I believe [Petro] got assurances that, you know, they would not be handcuffed, there would not be pictures taken, they would not be treated as criminals...” ([17:21])
• Scott Anderson: Warns of echo-chambered narratives, as both governments trumpet their version of events to save face. ([18:51])
• Use of Tariffs: Trump moved to weaponize tariffs (and other economic actions) under the International Emergency Economic Powers Act (IEEPA); while this is legally novel and possibly challengeable, it provides a fast way to exert pressure, bypassing normal regulatory and review processes.
  • Peter Harrell: “When he does use IEEPA to implement tariffs...there’s going to be some interesting litigation to sort out...” ([24:10])
  • Trump’s preference for tariffs over sanctions reflects both emotional and pragmatic reasons—tariffs can theoretically encourage reshoring, and he’s publicly criticized America’s overreliance on financial sanctions.

Memorable Quotes

• Scott Anderson: “Foreign policy by golf club and Truth Social.” ([10:32])
• Peter Harrell: “Trump...wants to send a signal...that diplomatic process you were used to...that’s over. If you cross the United States, I’m going to unload on you very, very heavily.” ([13:34])
• Kevin Frazier: On China’s comparative narrative: “The PRC has really leaned into offering stability...while Trump is instead focusing on short-term negotiations.” ([30:42])
• Eugenia Lostry: “There seems to be an opening window...that China is going to really exploit if they’re savvy.” ([33:07])

---

2. “Project Stargate” & Disruption in the AI Race

[33:47–59:23]

Key Points

• Project Stargate: Major U.S. government-backed initiative—a half-trillion-dollar partnership with OpenAI, Oracle, SoftBank, to build AI infrastructure, focused on “compute” (processing power).
• The Twist: Days later, China’s Deepseek unveils a highly efficient, consumer-facing AI model, releasing model weights (open-source), which matches top U.S. AI performance with radically lower resource use. This triggers massive financial panic—markets reevaluate if “compute” (hardware) is really the strategic moat the U.S. assumes.
• Five "Ingredients" of AI Power: Compute, data, algorithms, talent, energy. Deepseek's advance shows algorithms and talent might matter more than compute alone.
• Strategic Re-think: The U.S. bet has been on maintaining a “qualitative edge” at the frontier of AI (“AGI”) by cornering compute, but China’s approach—algorithmic and open—may accelerate diffusion and erode those advantages.

Insights and Analysis

• Kevin Frazier: “What Deepseek showed in releasing this far more efficient...model...is that maybe compute isn’t as important as we thought, and as a matter of fact, algorithms and talent...can go a lot further than we previously acknowledged.” ([34:43])
  • Stargate’s vision may already be undermined: the calculus for U.S. dominance needs revision.
• Scott Anderson: “Obviously Project Stargate was intended as a big investment in the energy and compute kind of inputs...where is the comparative advantage for the United States if compute’s less important than we think it is?” ([40:49])
• Peter Harrell: Noted that even Deepseek’s "low-compute" claim needs scrutiny—it was likely trained with substantial Nvidia GPU resources bought pre-export control.
  • The open-source posture of Chinese AI may prove strategically brilliant: “It does seem to be quite cheap to use...and I think China posting this...is an interesting play for developing world developers.” ([48:01])
• Kevin Frazier: Warns that U.S. export controls might be a “moat” easily bypassed if open-source algorithms close the gap.

Memorable Quotes

• Peter Harrell (on Deepseek’s open-source release): “This may be a very attractive model for you [as a developer in the developing world]—it’s almost as effective as the latest from OpenAI, it’s open source, and it’s cheaper to use.” ([49:27])
• Kevin Frazier: “If the rest of the world is going the open source route, then all these efforts to try to create artificial moats for U.S. leaders aren’t going to last very long.” ([53:28])
• Eugenia Lostry: “I don’t truly know how to square that with Trump’s comments yesterday that he is thinking about imposing almost 100% tariffs on chips and semiconductors coming from Taiwan.” ([55:22])

---

3. A 90-Day Freeze on U.S. Foreign Assistance

[59:23–72:24]

Key Points

• Trump’s Freeze: An immediate, near-total halt on U.S. foreign assistance for 90 days. Only a few programs were exempt (some food assistance, Egypt/Israel security aid). Dozens of USAID employees abruptly suspended.
• Global Impact: The U.S. provides roughly 40% of global bilateral assistance; sudden cuts risk collapsing critical programs—health, security, development—and destabilizing fragile partnerships.
• Programmatic Chaos: Many implementing partners operate with razor-thin margins, and losing funding even temporarily may permanently cripple capacity to deliver assistance or respond to crises.
• Foreign Policy Consequences: Foreign assistance serves strategic U.S. interests—health, security, development, cyber partnerships, and diplomatic leverage—often in competition with China’s own infrastructure and investment outreach.

Insights and Analysis

• Peter Harrell: “I have—I’m unfamiliar...with anything remote approaching the scale of this pause...I think that’s going to have some real world ramifications if they can’t sort this out and get at least most of it turned on quickly.” ([62:19])
  • U.S. assistance is about both humanitarian impulse and building/maintaining key alliances; both are jeopardized.
• Scott Anderson: “What a delicate ecosystem it is...when you pull a lot of this out, you are potentially really undermining the infrastructure you rely on to implement these programs.” ([64:44])
• Eugenia Lostry: Focuses on the cyber dimension—recent U.S. investments in international cyber capacity-building (e.g., response teams in Costa Rica) are now frozen, undermining years of efforts to win partners and counter adversaries.
  • “This was supposed to be a way for the United States to engage with countries that were maybe engaging with China...and now all of that has stopped.” ([67:22])
• Kevin Frazier: Points to demographic and strategic ripple effects, especially in Africa: “If we’re not actively investing in developing countries...that has huge geopolitical ramifications over the next decades as we see demographic changes accelerate in Africa and very much slow down in the rest of the world.” ([70:19])

Memorable Quotes

• Eugenia Lostry: “China was on hold, you know, like they are already there, right?” ([70:13])—underscoring how quickly China can swoop in to occupy vacuums left by the U.S.
• Scott Anderson: On potential for bipartisan pushback: “Ninety days is a long time to pause these indefinitely...Maybe we’ll see a correction on this path if we see a bipartisan reaction, but we’ll have to wait and see.” ([71:02])

---

Notable Quotes and Memorable Moments

• Eugenia’s Farewell:
  • “Don’t cry for me, Rational Security.” ([03:33])
• Scott (riffing): “Foreign policy by golf club and Truth Social.” ([10:32])
• Humor Over “Bazookas” and “Sticky Notes”:
  • Kevin (on Trump’s style): “Can we not bring out phones, let alone economic bazookas on the fourth green?” ([10:58])
  • Opening banter about minimalist “art” in remote recording spaces—“Optimized for art. That’s great.” ([01:33])

---

Timestamps for Key Segments

• Colombia Spat / Tariffs: 05:01–33:47
• AI Competition / Project Stargate / Deepseek: 33:47–59:23
• Foreign Assistance Freeze: 59:23–72:24
• Object Lessons (Fun Recommendations): 72:25–end

---

Object Lessons & Farewell to Eugenia

[72:25–79:08]

• Kevin Frazier: Applauded legal education efforts around AI, emphasizing “knowledge diffusion” and tech literacy. ([72:25])
• Eugenia Lostry: Signed off with a recommendation for the indie RPG “Wildermyth”—“it’s adorable,” a storytelling game where characters evolve across adventures. ([74:20])
• Scott Anderson: Recommended the solo journaling RPG “Thousand Year Old Vampire”—for those seeking escapes “when the real world gets a little too much.” ([75:49])
• Peter Harrell: Suggested the book “Freedom’s Forge,” about U.S. industrial policy and WWII mobilization, with lessons for current policy debates. ([77:16])

---

Tonal Note

As always, the episode balances deep policy analysis with signature Lawfare humor, anecdote, and warmth—especially for Eugenia’s last appearance. While topics are weighty and timely, the back-and-forth remains sharp, collegial, and insightful throughout.

---

Summary Conclusion

This episode offers a front-row seat to the abrupt—and at times, erratic—turns of U.S. national security and foreign policy at the start of Trump’s second term. From diplomatic standoffs and the weaponization of tariffs, to reshaping the technological arms race in AI and the risks of rolling back decades of global development investment, each segment explores not just what’s happening, but how, and why it matters for America’s standing and alliances. China’s steady hand and long-term strategy—in diplomacy as well as tech—underlies every debate. The conversation closes with a reminder of the importance of knowledge, adaptability, and perhaps an occasional “cute” RPG, in facing a turbulent world.

---

For more, listen to the full episode or visit Lawfare.

Transcript

[Marc Maron]

Hey folks, it's Marc Maron from wtf. Today I want to talk to you about Boost Mobile offering reliable nationwide coverage backed by a 30 day money back guarantee. Love your service or get your money back, no questions asked. Boost Mobile offers the coverage, network speed and service you're used to, but at more affordable prices. Why pay more if you don't have to? You can get an unlimited plan for $25 a month that will never increase in price and ever. No price hikes, no multi line requirements, no stress. Visit your nearest Boost Mobile store or find them [e-mail] After 30 gigabytes, customers may experience slower speeds. Customers will pay $25 per month as long as they remain active on the Boost Unlimited plan.

[Scott R. Anderson]

Eugenia listeners won't know this, but you have a stark and yet aesthetically appealing new recording backdrop. You're either in a very high end museum of some sort with one very small piece of art on the wall, which appears to be a blue sticky note on an otherwise blank white wall, or you are in the state of changing your furniture and moving things around a little bit.

[Eugenia Lastory]

Well, both are true in a way. You know, recently moved over the weekend, so everything is still in boxes. But I've also decided that in lieu of art, we're just going to put sticky notes that say what the art should be on the walls. So that's what you're seeing.

[Scott R. Anderson]

It's very tech sector. I like that.

[Kevin Frazier]

Optimized for art. That's great.

[Eugenia Lastory]

Yeah, yeah. Very efficient of us. Yeah.

[Scott R. Anderson]

Did you guys move across town? Are you in a whole different, like, neighborhood now?

[Eugenia Lastory]

No, we moved within the building. So we moved from the eighth floor to the fifth floor. And it was honestly the worst move I've ever had to do. 11 hours.

[Kevin Frazier]

11. I just. I top you. I'm sorry, I top you. So we moved from Helena, Montana to Miami, Florida. It took our movers five weeks to make that drive. When they finally arrived, what shoe wear were the movers wearing? Sandals. They were wearing sandals. And so every time they got up to the ninth floor to come and trolley out our stuff, they moved at a snail's pace. And guess who was charging by the hour? Our movers in sandals. And let me just add insult to injury, the trolley had a flat wheel. So not only were they in sandals, they had this flawed trolley. It was if I was ever going to be sent to jail for assault, it was at that moment my wife had to lock me in a closet.

[Scott R. Anderson]

That is more of a Miami problem than a move problem.

[Eugenia Lastory]

I think Foreign.

[Scott R. Anderson]

And welcome back to Rational Security, the show where we invite you to join members of the Lawfare team and our, occasionally our friends and associates as we try to make sense of the week's biggest national security news stories. I am your host, Scott R. Anderson. Thrilled to be back with you again this week with my colleagues, Kevin Frazier, our Tarbell fellow here at Lawfair. Kevin, thank you for coming back on the podcast.

[Kevin Frazier]

Always a pleasure, good sir.

[Scott R. Anderson]

And here again in a melancholy mood, or at least putting me in a melancholy mood because in fact, her last episode as she is sadly moving on to greener pastures in the near future. Eugenia Lostry, our senior editor. Eugenye, thank you for coming on the podcast, gracing us one last time with your presence. Although we're sad that we will be.

[Eugenia Lastory]

Losing you so soon, don't cry for me. Rational Security.

[Scott R. Anderson]

Well played. I was waiting for that line to come out.

[Eugenia Lastory]

Thank you. I've been thinking about it. Thank you. I appreciate it. I've been thinking about it since you invited me on. I'm very sad that this is my last episode, but I'm as always, glad to be on.

[Scott R. Anderson]

I like to think that every Argentinian person is just waiting for that one perfect moment to break out. That line.

[Eugenia Lastory]

No, I don't know. I don't think I ever thought about that line as much as after I moved here.

[Scott R. Anderson]

Yes, fair enough, fair enough.

[Eugenia Lastory]

I feel like it's an American thing. We have other things back home.

[Scott R. Anderson]

Yeah, it's more of a maybe a Madonna thing than Argentinian thing. Yeah, fair enough, fair enough. And joining us for the first time on the podcast, we're thrilled to be joined by Peter Harrell, a Lawfare contributor from time to time. I think he's been on one of our other podcasts, but never on Rational Security in the past. Former senior director at the National Security Council under President Biden, among numerous other roles in government, and now kind of a man about town handling sanctions issues, economic statecraft, and many other issues of which Lawfare listeners are very interested. Peter, thanks so much for joining us here today.

[Peter Harrell]

Yeah, thanks for having me on. I've been a longtime fan of the POD and so great to be able to join you guys.

[Scott R. Anderson]

Excellent. Well, we appreciate when people are willing to tip their toe in the more chaotic ends of the Lawfare greater cinematic universe, of which we are pretty much the sole exclusive product. So thrilled to have you on board as we talk through what has been a very news filled week and tackle a couple of the stories that have been occupying our minds and Our headlines over the past few days. Our first topic for today, tariff or takeoff. The Trump administration got into what is arguably its first major international spat this week when Colombia refused to accept a US Military flight returning migrants to that country, which led President Trump to threaten an array of punitive measures, from visa cutoffs to sanctions to tariffs. After Colombian President Gustavo Petro backed down, the White House was quick to claim victory. But how sustainable is Trump's strategy? And is it really the route to restoring respect for the United States that the White House claims it is? Topic 2 Talk to me when they get to project the Fifth Element last week, the Trump administration announced Project Stargate, an initiative not to revive 90s sci fi classics, but to instead make a massive investment in the US Development of artificial intelligence and related technologies. But a few days later, an announcement by Chinese AI platform Deepseek indicating it had reached comparable results to leading AI products at a lower cost than many of them in terms of resources, triggered a sudden decline in the value of AI related stocks and a broader panic. The AI bubble was bursting. What do these developments tell us about the competitive dynamics surrounding AI and how should the United States be navigating? And topic 3 a friend in need is a friend. Shit out of luck. The Trump administration has issued an across the board freeze of US foreign assistance programs for 90 days as it reviews them for consistency with the administration's vision of America first foreign policy. But what ramifications will this pause really have for US Foreign policy and beyond? So for our first topic, it was, I think, a pretty dramatic weekend in the international diplomacy front. Rarely do we see a spat, which seems like a pretty significant spat, spin up so quickly and then get resolved simultaneously. So quickly. And that I think if you stayed off social media wisely with your Sunday night or Saturday night, you may have missed the entire thing and only caught it in hindsight, which I think a few people very well may be in that position. But to review the basic details, of course, the Trump administration has started removing migrants from the country and returning them to other countries in the region using in part military flights. We've heard stories that at least some flights have been rejected from Mexico as well, denied landing access, although the Trump administration, at least last time I checked, was rejecting those stories, saying they weren't true. And then we had this incident where we saw the Colombian government, in particular the president, Gustavo Petro, deny this flight and actually follow up with a pretty, pretty damning public statement which, if you read in translation, essentially implied strongly that the Trump administration was acting in a racist manner and disrespecting Colombians, acting in a kind of colonialist manner and pulling in a bunch of different threads. This kind of long wandering but ultimately quite angry sounding statements. Nonetheless, in spite of that, President Trump followed up with a threat of an actual implementation of a whole array of punitive measures, visa actions, IPA sanctions, tariffs, all sorts of things across the board. And we quickly saw Petro back down and apparently concede to the point that he actually retweeted the White House's announcement summary of the call, which they said he had conceded basically all these sorts of points. It's a really notable incident a lot of ways. Cuz for a lot of people are saying this is particularly in the White House and President Trump's camp. This is the first of Trump's America first foreign policy really working. The White House essentially said finally we are, you know, America is respected again in the region on account of these actions. Peter, I want to turn to you first on this. I want to get into some of the technical nitty gritties of some of the things Trump threatened to do. But before I do that, you know, you obviously were in the White House, you've dealt with international incidents and diplomatic disputes to various extents in various contexts. Probably nothing quite like this one, although I'm not sure there was anything quite like this one. How unusual did this strike you as an incident kind of on both sides to some extent. And how much of a lesson do you think we should be pulling out of it, both for US Policy and maybe for the rest of the world about how to engage with the Trump administration?

[Peter Harrell]

Well, Scott, I guess I'd begin by saying we are definitely back to the era of foreign policy by social media post. And I think that we are all as observers here, just going to have to get used to this being the way things play play out over the next couple of years. But certainly was a stark, a stark reminder of that. You know, before we get into what exactly Trump threatened and you know, what could he have actually done and what maybe could he not have actually done, One of the things that jumped out to me putting on a bigger picture national security ad is actually these kinds of flights are quite routine, right? There have been flights to Columbia over the past several of several years. I believe there were some just over the last couple of weeks prior to this. And clearly the, the Trump administration was blindsided by the fact that shifting from civilian marked airplanes, bringing these migrants back to Colombia to military gray tales was going to provoke the Colombians and get them to Deny, deny entry. So it sort of shows. It's very unusual because usually staff have kind of handled all these kind of details and smooth things over with the foreign government before a plane gets into the, into the air. And I think kind of that part of this might show a little bit of the need for Trump's State Department to settle in and DOD to settle in and get to kind of ordinary course engagements, because that was quite surprising. And then of course, Trump, faced with a plane that might not be able to land, kind of pulled out the economic bazooka and pointed it at Columbia over the course of a couple of hours, apparently while he was out on the golf course. So, you know, multitasking Sunday afternoon, I.

[Scott R. Anderson]

Hope his lawyer was with him, advising him on how feasible a lot of the threats that he provided were. Exactly. Because it was really quite arranged, which is quite impressive.

[Peter Harrell]

Yeah, I'm not convinced he was golfing with his lawyer. I'm also not convinced that White House Counsel's office reviews his social media post stream. But I've never served in the Trump administration, so I can't, you know, actually speak to how much legal review, the, the truths, I think he calls them on truth social actually get.

[Kevin Frazier]

I gotta say though, that is horrible golf etiquette, though. I mean, can we not bring out phones, let alone economic bazookas on the fourth green? Like, let's just leave the phone in the bag.

[Scott R. Anderson]

So this incident, whole thing really struck me as something that really spiraled out of control in a couple of different directions. Like I actually sympathize. I think the reason you noted, Peter, a little bit with the Trump administration in that this strikes me as an initially a bad tack for the Colombian government to take. Really. Like they obviously took an incident and Petro, as I understand it, it not being somebody who follows Colombian politics very closely, but I understand it elected a little bit on a kind of left leaning, not anti American, but reservations about perceived American imperialism, critique of aspects of the US Colombian relationship sort of platform that's been part of his appeal. And he's kind of flagging the polls maybe either out of genuine good faith or as an effort to stir up a political issue. Obviously took this as great umbrage, something that might have seemed relatively routine. And then of course, Trump kind of responded in kind. Right. Like one approach you could have seen the Trump administration do is say, okay, what we care about is get these migrants back to the country. We'll just put them on a civilian flight and then we won't have to deal with this. Issue again. But instead you saw the bazooka come out and say, no, there's going to be military flights. And that's significant to Trump because he's made military being involved in immigration enforcement a major platform of his, and it's not something that he wants people pushing back on, particularly international audiences. That's at least my read of the dynamics. Is there a lesson there, though, about what the trade offs are for the Trump administration in employing these sorts of tools and that sort of threat, valuing that symbolic significance, which I guess is more for a domestic audience more than anything, I guess, from your perspective in the Biden administration elsewhere, Peter, what would the trade offs be? Is there a reason that you might have taken a different tack or advised a different tack when you were in government? Because there are costs associated with such an aggressive response?

[Peter Harrell]

Well, look, I think in general, you'd see most administrations take a pretty different, in some cases, potentially even a little bit excessively bureaucratic approach to this kind of thing, need an interagency meeting on a Sunday afternoon, write a memo on what the options are, and then maybe by Wednesday we would have figured out a couple of recommendations for our betters to consider. So I do take the point that the ordinary course of diplomacy can probably move a little too slowly here. That said, there has to be some reasonable middle ground between that and what we saw play out on Sunday afternoon. I actually think Trump was doing this and kind of unloading very heavily or threatening to unload very heavily on Colombia. Uh, yes, as you say, Scott, partly for his domestic audience, he wants to show, I am repatriating these migrants via military flights. This is a top priority for me. I'm getting my way. I do think there was a domestic element of it. I also think there was very much an international diplomatic element of this, where Trump is coming in and he wants to send a signal to foreign, foreign leaders that, you know, hey, that, that, that diplomatic process you were used to of kind of niceties and, you know, U.S. government kind of debating what to do that's over if you cross the United States of America, I'm going to unload on you very, very heavily. And I think he's thinking about that, particularly as he gets into trade negotiations with a variety of countries around the world, likely over the coming weeks and months. And also as he looks at other issues like, you know, potential migrant repatriation flights, flights elsewhere, I think the downside of this is that, well, you know, well, in the short run, he does seem to have gotten what he wanted. I Mean, the Colombian seemed to have caved pretty quickly. You know, I'm, I'm a strong believer that over the long run, particularly in this kind of geopolitically divided world we lived, we live in, it's important to have allies and partners on your side. You know, folks are willing to work with you on, you know, issues around China, on, you know, Columbia's got a lot of critical minerals and things like, on issues around supply chain resiliency and, you know, if you tick these guys off too much, well, they might cave to you. In the short run, I do think you run some risks that over the mid and longer run, your partners may drift away from you. So I think that's the kind of downside. Short run, I see why, you know, he probably is of the view he got what he wanted, but you'll have to see what the, the longer term ramifications here are.

[Scott R. Anderson]

Let me actually come to you on that, Eugenia, because I'm kind of curious. You have a kind of unique perspective and experience you can bring to bear on this, although from slightly different context. I don't want to exaggerate too much the direct relevance here, but you worked in a foreign government, in foreign affairs and diplomacy at an earlier stage of your career for the government of Argentina. And I should say, I think during, at least overlapping a little bit with Trump's first time in office. I'd be kind of curious about how you think, at least in your experience, the Colombian government or other governments in the region, other foreign governments kind of approach the United States generally, and particularly the Trump administration perhaps specifically in these sorts of moments. Is my instinct, right, that Petro's sort of response on this was a little bit of an outlier or unusual or kind of a high risk strategy, something that other governments might avoid for this sort of incident. Or is he onto something? Is it reflecting a strain of thought about how to approach these sorts of scenarios?

[Eugenia Lastory]

So I have to admit that I'm still kind of a little bit confused about the order of events here. Right. You mentioned that Trump got what he wanted, that Petro ended up giving up whatever Trump wanted. But as far as I understand it, some of the concerns that the Colombian government had about these flights was that they were coming in military flights. And given the experience with some Brazilians that were being returned to Brazil, I think a day before that, where they were handcuffed, there were pictures taken of them, they were not allowed to use the bathroom during the flights, the concern was that the dignity of the detainees that were being returned to the country was not being respected. And that was fundamentally what the Petro administration was complaining about. I don't know that that was exactly clear from the very long post on X that he had. That was a little bit confusing and I think it touched on, on many things. But I think we saw similar comments coming from, from Mexico, for example, and Brazil about, again, the dignity of these people not being respected. And so I think you could also make the argument that in a way, Petro also got what he wanted because I believe he got assurances that, you know, they would not be handcuffed, there would not be pictures taken, they would not be treated as criminals, and that they would, you know, be allowed to use the bathroom and then they would return to the country. So probably you're not going to see many other, especially after the bazooka that, that Trump pulled out based on the response. I don't think you're going to see too many other posts on X from foreign precedents saying, you know, you know, we don't like the US or we don't shake hands with white enslavers. But I, I do think that there is something to what the complaint was about and that that was received in the Trump administration and it was handled. And now, you know, the, the flights have resumed and some other countries will maybe not put up that much of a fight, but they are indicating that they will receive these military flights.

[Scott R. Anderson]

Yeah, and that's actually a really good astute point that's worth flagging here, which I think is actually a part of more often a part of the Trump diplomatic calculus than that doesn't get as much attention in the media in that, you know, I don't know whether this is Trump being a bargainer and a businessman or whether just the reality of global diplomacy, but he actually is somebody who makes concessions. Like he's willing to strike a deal and make little hand me down things, usually trumpets a deal as if he has vanquished an enemy, bullied somebody into doing something. That was undoubtedly the case here. And again, it's really interesting that Petro essentially retweeted the White House summary that said, I think almost verbatim, Colombia accepted all the United States conditions, retweeted it, seemed to accept that as a condition. But then in his own commentary and press from his own government said essentially the United States actually did acquiesce on these points are most important to us, it'll be military flight, but the standards of treatment are going to be better. We don't know 100%. What this is the problem with these sorts of bilateral agreements is that you get only two people in the room and all you can take is their two narratives, which don't always meet in the middle. I suspect it is somewhere between the two to some extent. And so that that is a little bit of a safe phasing measure and a part of the equation here that even though the Trump administration isn't trumpeting it or even really acknowledging it publicly, very well may have entered in here. We know this was part of negotiations around prior the remain in Mexico project and stuff like that. There were kind of concessions made along the line to get to. Yes. And so that kind of act of bargaining is part of the equation and something that to definitely consider here. Let me turn now, before we run out of time on this topic, to this more technical question, because something that's really notable about this incident is that it's the first time we saw Trump pull out one of his big tools, his big sticks that he's been talking about since he was reelected, but since before he entered office. And that is this question of tariffs. And I think we actually have, according to press accounts, statements that his White House had drafted executive orders. I think actually I saw media specifically saying they're under the international words, the Economic Powers act, which is kind of atypical tariff authority, saying that he was ready to impose these things and was keeping them drafted in case Petro went back on the terms of this agreement. And there was also some visa restrictions that actually were implemented and are still implemented until the first flight lands, at least according to the White House. Peter, talk to us a little about these authorities. Like, you know, how credible a threat is this? Because this use of IPA as tariffs, it's something we saw Trump threaten to do and implement for Mexico briefly in his prior stint in the White House, but never actually fully implement, never actually apply. As I recall, it talks to us about how feasible this tool is and the other tools that we saw the president break out in, like how credible a threat is it he's actually able to bring to bear in these sorts of scenarios.

[Peter Harrell]

Yeah, so that's a, that's a great and really important question. I think that even though the kind of, you know, current tariff spat seemed to get resolved Sunday, Sunday evening, it seems very clear that we are going to have plenty more opportunities in the coming weeks to get back to this discussion and to dust off what we think of the tools Trump can use here to impose, impose tariffs. So I EAPA the International Emergency Economic Powers act is a law that Congress passed back in 1977 basically to give the President a broad range of mostly economic authorities in cases that the President determines are an emergency. And it's been used many, many times by president, starting with President Carter during the Iran hostage crisis to impose sanctions, so, you know, freezing the assets of our adversaries and also imposing various kinds of trade embargoes with adversaries. President Biden, for example, used it in 2022 and since for US sanctions on Russia, which have included banning certain categories imports from Russia. And President Volta used it for various other things for export control purposes. It's currently being used for some of these information communications technology restrictions the Biden and Trump administrations have put in as well. So it's a pretty sweeping statute, but in all that time since 1977, it's never been used for tariffs. And that reflects in part the fact presidents have lots of other authorities that they could use for tariffs. And it reflects in part the fact some ambiguity about whether IPA can be used for tariffs. I, I, I think the reason Trump would like to use IPA for tariffs is that if he's allowed to use it for tariffs, there's almost no procedural oversight of it. He can kind of sign an executive order, slap the tariffs on, they can come into course almost immediately. Whereas all of these other trade statutes that Congress has enacted over the years, like Section 301 of the Trade act in 1974, that's the statute President Trump used for his trade war with China during his first term. If you want to impose tariffs under, under section 301, USTR has to take a, like a six month investing, six month plus investigation. They have to propose tariffs, there's notice and comment process. So, so these other statutes just bring a lot of process. And, and, and Trump, I think it's safe to say, has never been real big on bureaucratic process. And so, and he's also has a lot of ambitions on tariffs. And so it kind of makes sense from his perspective that IPA, where you can, in theory at least, just, you know, take the sharpie to the executive order and sign Donald J. Trump there at the bottom and the tariffs come into force, force, that that's what he'd like to rely on. I think that when he does use IIPA to implement tariffs, and I think that it is going to be a question of when and not if. Because if I look at all his tariff threats, I think AIPA tariffs are coming. There's going to be some interesting litigation to sort out, you know, whether, and to what extent AIPA does authorize him to use, to use tariffs. And well, you know, we can get into this in detail, but while I think a court would probably uphold it vis a vis Columbia, there's certainly lots of arguments that one could make on the other side. And, you know, we'll have to see where the courts come, come down on that. You know, final point to make, Scott, is just, you know, if the courts didn't let him do it under aipa, he could still find another way to tariff Columbia. It just that wouldn't come into place until November and that's not the kind of threat he wanted to make.

[Scott R. Anderson]

Yeah. And actually leads to one other question which I'd be curious about, about your take on, which is why is tariff the tool of choice for Trump? Right. Like every other recent president, they've got a whole range of tools and economic statecraft tools are like one subset of those. It's become more prominent. But sanctions is the kind of most common tool we see most presidents impose in a variety of international situations. Now, usually it's like terrorism, narco terrorists, instability in a region, global corruption, human rights abuses, other things we've seen sanctions applied for. But you could see, I think you can make a case of a, you know, an IPA based sanctions regime. Like we see a lot of other places to target Colombia. If you wanted to do that, it entail, I guess, targeting specific industry actors, maybe a little more complicated implementation than an across the board tariff. But is there a reason why, if nothing else like it seems like that would let you carve things out to actually reduce the domestic ramifications, which is always like a constraint on the president's ability to credibly threaten this is that it actually is going to be felt by American consumers and that's going to have political ramifications, among other things. So why is it, do we think that is there a big competitive edge that tariffs offer over sanctions that he keeps turning to it, or is it part of his worldview in that it feeds into his sense about the role that trade plays more generally as a situation where the United States is kind of disadvantaged and there's two birds with one stone approach in using tariffs for these sort of scenarios?

[Peter Harrell]

I think this is a really interesting question, and I should caveat this by, by saying I think it can be difficult to opine into kind of the inner mind of Donald Trump and what exactly he's he's thinking here. That said, he has actually talked about this issue to some degree. And the first point I should make is he actually threatened a bunch of sanctions under IPA in addition to the tariffs. The sanctions were kind of going to come into place later than the tariffs. And I think it would be legally almost certain that he could impose sanctions on some Colombian government ministry for refusing to let these planes land. The only open legal question would really be about the tariff authority. So in some ways, had he chosen sanctions, it not only would have been more consistent with what presidents going back almost 50 years have done, it also would have just eliminated any legal ambiguity or almost eliminated any legal ambiguity here. So I think, you know, against that backdrop, you have to sort of look at this as Trump, almost more than any other American political figure in the last, you know, at least in my adult lifetime, really is attached to tariffs. You know, he talks about that, right? He's a tariff man, and tariff is the most beautiful word in the English language. And I think he really likes tariffs over these other tools for kind of two reasons. First, he does seem to think that tariffs, unlike sanctions, can help reshore, you know, that industry here to the United States. So, you know, maybe he was hoping if he tariffs, you know, for example, flowers from Columbia, we import a lot of flowers from Columbia, maybe it would help, you know, rose growers in Miami, something like that. Right. Maybe he thought there'd be a reshoring play that he could get with these tariffs in addition to the. The economic coercion of. Of Columbia. Uh, and then he also talked on the campaign trail. This is not something he talked about during his first term, but he's definitely talked over the last 18 months that he thinks we are using sanctions too much, and he thinks US Financial sanctions are driving folks away from the dollar, which is interesting, like a concern a lot of sanctions experts has had. Um, but Trump has said on the campaign trail he thinks, you know, tariffs are the solution, that of it both gives economic leverage over these countries, but doesn't really encourage them to move away from the dollar. So he does seem to have, in addition to, you know, what I might characterize as a very strong emotional attachment to tariffs, he does also seem to have some kind of intellectual idea here that over the long run, tariffs will be more beneficial to U.S. interests than what he sees as a very heavy reliance on sanctions.

[Scott R. Anderson]

That's really interesting. That's something I want to dig into deeper, maybe probably for a longer conversation than we have time for today. Let me turn to one other aspect of this before we move on to our second topic, and that is the competitive aspect of this. A lot of the media coverage has, I think, properly situated this tension, this conflict in the context of global power competition, particularly in relation to China? I think you saw another commentary. I think it was, it could have been from Senator Murphy. Apologies, Senator Murphy, if I'm misquoting you here on social media, noting that China was celebrating this sort of Trump response. I know the Chinese embassy in Colombia did release, at least I saw a quota in the media press report noting the strong relationship with Colombia and the fact that they're approaching, I think, a notable 60th anniversary or 70th anniversary of establishing relations, something to that effect. So, Kevin, let me turn to you. You thought about this global competition context in certain nexuses, technology, AI in particular, but just kind of as a general angle, how do you think this fits into the competitive narrative? What is the comparative advantage? China, maybe to some extent Russia, I think is really a China US Competition. What is the narrative that gives them a competitive advantage that they think might give them a comparative advantage in appealing to other countries in the Western Hemisphere, among other places around the world? And does this play into it? Does this rebut this? Is this a lateral move that doesn't really impact that? What's the relationship between that geopolitical competition and these sorts of diplomatic maneuvers?

[Kevin Frazier]

Yeah, I think if we look more generally at what the CCP has prioritized and what the People's Republic of China has prioritized would be offering stability, which is maybe the exact opposite of a sort of negotiation with a President Trump, as we saw over the weekend. And so whether we're looking at the Belt and Road initiative or offers, for example, to help build out AI infrastructure, which I'm sure we'll get to soon. The People's Republic of China has really leaned into offering stability and economic growth and infrastructure investment, all of these things that are very appealing to the developing world in a way that Trump is instead really focusing on kind of short term negotiations. And so while that may benefit the US in the short run, the long term ramifications of who's going to move into whose sphere of influence I think is something we're going to have to watch over the long term countries that increasingly find themselves having to respond to posts on X and calls from the golf course maybe aren't going to be particularly keen to work more closely with the U.S. if we continue to see that sort of negotiation go on. So I wouldn't be surprised here if China's response was really to lean into some of those more long term initiatives that make folks want to move over into their camp based off of assurances of continued economic support and knowing that they're not going to face these sorts of random disruptions on a given weekend.

[Eugenia Lastory]

So few things. One, I do want to note that at least the Chinese ambassador to Colombia did note that the interview that he did, he did a week before this whole situation happened. And he was responding to some people who were saying, like, oh, are you taking advantage of this? And he was like, no, this interview was already planned. I did it a week before. I didn't really know that this was going to happen. But I do think that underscores Kevin's point, right, that there is a stability that they are likely going to take advantage of. The fact that they can just say, like, yeah, we're approaching 45 years of, like, good relationships and this is only going to get better. And a week after that there is this diplomatic crisis on Trump's kind of first week. I do think that there is, and I think this might be a threat that connects all of the issues that we're going to be talking about today. There seems to be an opening window, at least from what I see Trump doing, that China is going to really exploit if they're savvy. I think you can see presidents conceding to Trump. Trump may say we're respected again on the international arena, but I think that's kind of a short term perspective. I think if this is what you can expect relationship with the Trump administration to be, you are going to start looking at other options with kinder eyes in a way.

[Scott R. Anderson]

Well, I dare say I think we may have reason to revisit a little bit of this conversation for our third topic about foreign assistance. But before we get there, let's shift to our second topic about Project Stargate. My favorite, favorite reference and name I've ever brought out from the Techsecker. Palantir is great. Gotta love that. Every Lord of the Rings reference, I'll get it. But Stargate, you are really digging, digging deep into the Kurt Russell oeuvre on that one and I greatly appreciate it. Kevin, talk to us about what we've seen happen last week. We have two data points. We came into this segment actually just talking about Project Stargate, and then we saw this deep sea revelation come out. Or maybe it is, maybe it isn't revelation, but certainly the market seemed to be treating it that way. Talk to us about the push and pull of these two different factors because it's an interesting juxtaposition, juxtaposition of two pretty major developments in the AI space that are triggering such kind of divergent reactions.

[Kevin Frazier]

Yeah, the ground certainly fell out from under whatever expectations we had about the AI development race or the AI arms race going on between the US and China over the past week. All you have to do, or maybe don't do if you're overly invested in Nvidia right now, but if you look at your portfolio and you see half a trillion dollars wiped off of the stock market, or up to a trillion dollars for some tech stocks, this was definitely a huge week economically. What's interesting is that really close observers, this has been their time to shine on X because they're citing all their prior research. They're going back to that article they wrote in June where they say, hey, there's this company Deepseek, watch out, they might be relevant, we should keep an eye on them. Well, sure enough, those folks have really, really made the most of that call. And we saw that came to fruition just recently when Deepseek introduced a consumer facing model for the first time that was far more efficient, involved far fewer resources than OpenAI competitors or anthropic competitors, and showed that maybe we don't need the sort of incredible amount of investment in computer that OpenAI has been calling for. And yet that's exactly what we saw just about a week ago when in the White House, President Trump was joined by Sam Altman of OpenAI as well as Oracle and SoftBank representatives in the announcement of the Stargate project. The Stargate project, in short, is a half a trillion dollar bet on AI infrastructure. And that's supposed to occur over the next four years, creating as many as 100,000 jobs. Thanks to some good reporting from Bloomberg Law, we know that 57 jobs have been created so far. So we have 99,943 to go. The upshot of this project is really making sure that OpenAI specifically has the AI infrastructure it needs to continue to build and, and train frontier models. And this is operating off of the assumption that of the core ingredients of AI, of which there are five, which I'll get to in a second, compute is one of the most important ones. And that computational power to train models and to be able to use models is probably the most important ingredient. Well, there are five total ingredients. We have compute, but we also have data, algorithms, talent and energy. And what Deep Seq showed in releasing this far more efficient open source model and by open source here, just to appease my friends who are open source diehards, they released the model weights, not full open source, but this open source model. They showed that maybe compute isn't as important as we thought, and that as a matter of fact, algorithms really and talent in particular can go a lot further than we previously acknowledged towards releasing some of these frontier models. And as a result of showing that maybe compute isn't the end all be all, we saw quite the disruption to what folks were hoping the Stargate project would unleash in terms of realizing President Trump's goal of AI global dominance here in the U.S. hey folks.

[Scott R. Anderson]

Scott R. Andersen here. Things are finally beginning to cool down here in Washington, dc. There's a nip in the air. The days are getting shorter. Football is back on tv. All the signs are here. Autumn is coming. While the kids are sad to be back in school, it's a favorite season for those of us who are bearded, burly and sartorially inclined, as it's finally a chance to bring back the flannels, sweaters, denim and tweed we had to put away for the summer. But if you need a little help getting your closet ready for the season, don't sweat it. Quince has you covered. From new denim jeans to leather jackets to woolen sweaters, Quince has everything you need for cooler days ahead. And all of it is comfortable, classic and high quality. The types of items that will stay in rotation for years to come. The Mongolian cashmere sweaters, which go for an incredibly reasonable 60 bucks, are what got me into quints in the first place, and a few are going to be back in my regular rotation this fall. I also picked up an Italian wool overshirt that's set to be my main light layer when things get nippy, and an organic cotton long sleeve polo that's already begun to make its way back into my regular outfits. Plus, Quince is now offering more than just clothes. From houseware to luggage to rugs, Quince has tons of the everyday items you need. Personally, I'm thinking some flannel sheets may be just the thing to get before the temperatures really drop. And the best part is that by cutting out the middlemen, Quince offers all these high quality products at half of what they would cost you at other stores. Plus, Quince only works with factories that use safe, ethical and responsible manufacturing practices. So keep it classic and cool this fall with long lasting staples from quince. Go to quince.comsecurity for free shipping on your order and 365 day returns. That's Q U I N C E.com Security to get free shipping and 365 day returns. Again, that's quince.com Security now let's get back to the show.

[Marc Maron]

Hey folks, it's Marc Maron from wtf. Today I want to talk to you about Boost Mobile offering reliable nationwide coverage backed by a 30 day money back guarantee. Love your service or get your money back, no questions asked. Boost Mobile offers the coverage, network speed and service you're used to, but at more affordable prices. Why pay more if you don't have to? You can get an unlimited plan for $25 a month that will never increase in price, ever. No price hikes, no multiline requirements, no stress. Visit your nearest Boost Mobile store or find them online@boost mobile.com After 30 gigabytes, customers may experience slower speeds. Customers will $25 per month as long as they remain active on the Boost Unlimited plan.

[Scott R. Anderson]

And just to sharpen this kind of assessment for listeners who may not follow us, like what does that mean? Mapping onto China versus US Global competition? Like obviously Project Stargate was intended as a big investment in the energy and compute kind of inputs.

[Peter Harrell]

Right?

[Scott R. Anderson]

It's like about solving energy problems, locking down processor development and manufacturing. Right. As I understand it, maybe there's other parts of it too. But where is the comparative advantage for the United States versus China rely in this competitive dynamic? And so what does it mean if maybe compute's less important than we think it is?

[Kevin Frazier]

Yeah, so I guess one place to start would be to your point, Scott. There are a lot of details about the Stargate project that we don't know. There's an ongoing debate about whether SoftBank actually has the money it's pledged. If you want to look at the Twitter fight going on between Elon and representatives of the administration, you'll see an ongoing discussion about whether the $100 billion that was supposed to be invested in this first year is actually ready to go. In addition to not having a lot of certainty about the funding situation for the Stargate project, there's a lot of speculation about whether the plans have actually been outlined to make sure this build out occurs as quickly as possible. Some astute observers, for instance, noted that OpenAI had a form on its website. I haven't checked recently, but at one point they had a form saying, hey, if you're in construction and or land and or data center development, hit us up. Well, if you're trying to build something out within four years with half a trillion dollars, ideally you've got all the plans, all the contractors, all the construction folks lined up. This is a space. OpenAI isn't the only company looking to build out AI infrastructure. So there's a lot of details out there we're not sure we know for the Stargate project. So putting that to one side, I think the key thing that is causing a sort of panic among those really pushing for the US to dominate the AI space is the fact that in relying on computer as the main way to ensure a sort of moat for US AI companies against their peers in China and around the world, we may have actually induced China to become more advanced in those other ingredients with respect to frontier model development, so becoming more sophisticated with respect to algorithms and training on less data. DeepSeq importantly has a really fascinating background. It was initially spun up by a hedge fund, high flyer, and was focused on identifying good stock plays in the Chinese market. And so up until this point it hasn't had any consumer facing products. Why that matters is it was able to fly under the radar and get out of the way of a lot of Chinese regulations with respect to AI. And so the founder of Deep Seq has been very clear since they got started that their emphasis was going to be on research. And so they've just been plugging and plugging away on hiring diverse talent, expert talent, and fine tuning their ability to make the most of fewer resources than their competitors. And now we see that, huh, maybe this access to compute isn't as important as we thought. And all of that attention we spent with respect to export controls, for example, under the Biden administration, may not be as transformative, as substantive as we were hoping in ensuring US dominance in this space.

[Scott R. Anderson]

That's actually a perfect tee. Up to the question I want to ask you, Peter, which is looking back at both how the Biden administration and the broader US government has come to this. We don't know exactly what the Trump administration is going to do, but there's going to be in kind of the broader major power competition. There's plenty of threads of continuity between the way US government is going to think about this across the Biden and Trump administrations and from the first Trump administration, and to some extent going back to Obama. And there's going to be a core element of that, although I think that the valence on AI may vary a little bit in terms of different shades and perspectives. My sense of it, correct me if I'm wrong, is that the focus has really been for the past few years on the frontier AGI vision. This idea that we're going to keep a qualitative edge on a variety of fronts in this AI competition. That's going to be the US advantage It's all about pulling the ladder up, preventing people reach there. And that the compromise there has been on this diffusion question, like lower models that may not hit that optimal capacity but have broader global reach and that we see different countries tapping into because China is taking a little more of a horizontal breadth strategy as opposed to the pinnacle strategy, in part because they have to, because of export controls, or at least they're less advantageous in that position. Is that a fair way to characterize how this competitive dynamic has played out? And I'm curious as to whether this might be, or we might be approaching an inflection point where there seem to be a more complex set of trade offs between that frontier AGI versus diffusion versus the other variables that go into this competition.

[Peter Harrell]

Yeah. Scott, I think you broadly, accurately characterized one of the kind of core strategic concepts of the Biden administration's approach to AI competition with China. I think the Biden administration for four years, and it really thought about AI going back to the beginning of the administration, and there was some work that was done in the late Trump term one on this too. But there was this very strongly held view that the us you know, has a qualitative lead in AI that is very much in America's interest to maintain that lead, given how important AI is going to be to a whole, whole bunch of applications, both economic and military applications over the coming years. And that what we need to do is A, invest here at home in order to maintain that edge and B, as you say, kind of pull up the ladder on the Chinese. And there was this view that the most effective way to pull up the ladder was to limit China's access to advanced compute. Because the theory was that China, these AI models need lots of advanced compute to really develop and to train, to train on your high end semiconductors, the kind of things Nvidia's worth $3 trillion or was worth $3 trillion for making. And that's something you can actually control. You know, American firms are designing the semiconductors. You can actually keep the Chinese from buying them. You know, in theory, maybe you can control algorithms or something like that. But the Biden administration view of it's actually like quite hard to prevent data from leaking over to China. So the thing you can control is compute. And so you saw this really aggressive effort first to restrict China's access to advanced computing chips and then at the very tail end of the administration to kind of globalize these rules to say really the most advanced chip should be most available in the US and in close access allies and then when they're going to be in countries like in the Middle east and whatnot, they should be built out by kind of trusted western firms because we're going to, you know, want our guys to really lead in the development of this AI. And that's kind of a core strategic concept here. Now I think, you know, what we saw with Deep Sea coming out of China represents an interesting challenge to that. Now the first thing I should say is I think there are some real questions about, you know, is deep seeks model really as efficiently trained as they say it was? This hedge fund that Kevin talked about, it acquired 10,000 Nvidia GPUs before, back in 2022, before the export controls came in. So there is actually a fair amount of advanced compute that these guys had access to. And I think that that raises some question. You know, we shouldn't just sort of assume this is something that, you know, was trained on my kids remainder iPad had from five years ago, right? That's not actually true. So, so I think we got it, we got a pressure test what actually happened here that said it, it, you know, does by all accounts seem to be really interesting algorithmic developments that I think should cause the US government to, you know, test Is this assumption that compute is the way to control China's AI development, is that really the most effective vector? Should we be thinking about other vectors? Should there be other things that we are, we are doing here? You know, final point I'll make on this and it's interesting because you know, the, this Chinese model is open source, right? They've put it, they've put it online and it's, you know, whether or not it was as cheap as they say it was to develop, it does seem to be quite cheap to use, you know, for people running queries on it. And I think that China posting this, you know, allowing this to be kind of open source. I think probably Chinese firms talking up this, I think it's an interesting play China's making for kind of what developers in the developing world are going to do. Right? If you're a developer in Indonesia or in India or in Brazil, this may be a very attractive model for you, right? It's almost as effective as the Latest out of OpenAI AI. It's open source and it's cheaper to use. So I think they should also be getting US government to think about how do we keep up play and make sure our guys are out there competing for developing world developers. Final point I'll make just because I signed up for it over the weekend when the news was really developing before they shut off new American enrollments and, and, you know, with Howard Lutnick's confirmation as Commerce Secretary coming up, I decided I'd ask Deepseek a question to ask Mr. Lutnick at his hearing about strengthening semiconductor export controls on China. You know, what does Deep Sea think about strengthening semiconductor export controls on China? And it actually gave me quite an interesting response. It proposed that the Chinese Communist Party's military civil fusion Strategy directly exploits U.S. and allied semiconductor technology to modernize its warfighting capabilities from AI enabled weapons to surveillance systems. If confirmed, would you commit to dramatically exploring expanding controls, not just on advanced AI, but also cutting edge tools, semiconductor materials, and emerging technologies? So it seems to have a hawkish view on this question.

[Scott R. Anderson]

He just may be reading the, reading the tea leaves on that one. Politically at the moment. I feel like that's fascinating. Don't ask about Tiananmen Square, but that's a good answer that we can get out of.

[Peter Harrell]

No, to your point, if you ask anything directly about China, it sort of says, I can't talk about that, that let's play a game kind of language.

[Scott R. Anderson]

Oh, I love it. I love it.

[Eugenia Lastory]

So besides the, you know, interesting technical advances that Deep SEQ presents, one of the things that I was interested in when I saw some of the coverage of it is that there seems to be quite a difference in how transparent Deep SEQ has been compared to some of the American firms. So given I'm directing this to you, but maybe anyone who knows, you know, what are they actually transparent about and how does that compare to maybe the American way of talking about their advances?

[Kevin Frazier]

Yeah, I think one of the most obvious things is just that first, the open source mindset. So the open model weights is something that you can't get from OpenAI, you can't get from Anthropic. Interestingly, we may see another effort by Zuck to have to appease President Trump, because some people are saying that Meta's decision to open source its llama model may have had a direct role in helping Deepseek advance to where it is today. So keep your eyes on that relationship. But I think in addition to making their model open. Wait, what we've seen from Deepseek is a real willingness to publish in open formats. On Arxiv, for example, some of its innovations and advances in a way that we're not seeing its rivals in the US Do. In the same way, the leaders of Deep SEQ have been clear that they're really wanting to be a research outfit, and that they regard being lead innovators in an open way as a key mission of the company itself. To Peter's point, whether we see the PRC continue to embrace that is an open question. An expert at Rand posited that China didn't have Deep Seq on its five year plan. No one saw this necessarily coming. And so whether the PRC decides to continue to allow Deep Seq to operate in this fashion remains an open question. But I do think that the ongoing conversation now is that the US really has to change its approach to AI innovation, that if the rest of the world is going the open source route, then all of these efforts to try to create artificial moats for US leaders aren't going to last very long. Gary Marcus had a very interesting piece analyzing Deep Seq's advances in this space. He pointed out that ByteDance, for example, recently released its own model that was cheaper than R1 released by Deep Seq. And a Hong Kong lab released yet another similarly capable model as R1 with even less training data. So as we continue to see more and more actors pursue an open mentality to AI development, it's going to ask the question for US Leaders, how are we going to compete with everyone around the world if we're not trying to diffuse knowledge across more actors? And that's something that we haven't quite seen the administration embrace to this point, but may have to be a certain pivot in the future. And one final thing I just want to shout from the mountaintops. We're not talking enough about general AI literacy. It's very important that countries focus on frontier development. But knowledge doesn't mean a whole lot if it's not diffuse. And so the country that has the right pairing of AI innovation with knowledge diffusion, the public that can actually make the most of these tools, that's what you need. You need both of those. And we're not talking enough about that diffusion aspect. So that's one other arm of the transparency conversation that I think needs to be brought up more frequently.

[Eugenia Lastory]

That's great. Something that has come up a couple of times is that the US needs to rethink how they're thinking about innovation when it comes to AI. I don't truly know how to square that with Trump's comments yesterday that he is thinking about imposing almost 100% tariffs on chips and semiconductors coming from Taiwan. If you can make that make sense for me, that would be. That would be awesome. I don't understand it, actually.

[Scott R. Anderson]

One compliment to that question, Kevin, because in the same vein, if talent is a bigger input than we realize or more variable and scalable input, how does that integrate with diffusion? Is the diffusion of American based models and technology and access to the other inputs? I understand a lot of this is about US homing compute and energy resources and stuff like that, right? So if you're overseas, you can't access them as easily, although you still can to some extent. What is the trade off for talent? Are you really cutting off a talent pool? Because a lot of really talented people in the rest of the world who you might otherwise be able to access aren't going to be trained on US models how to use them, Are going to naturally gravitate towards Deep SEQ or something else? Or is there less of a trade off there than there might be in other technologies? If you're thinking about instead, I always draw comparison to Huawei and telecommunications competition, right? Like broad diffusion was part of China's strategy very deliberately there despite low performance. And that became a big concern down the road. But we don't seem concerned about it yet. Is that something we will be down the road more?

[Kevin Frazier]

Yeah. So to start with Yuhenni's question to be clear, as Peter pointed out, Deepsea didn't create this model with like one chip, right? They had 10,000 Nvidia A100 chips. And so compute still matters. And that's a very important aspect of this conversation is that having some advantage in computer is still an advantage and is something that's going to help the US develop more sophisticated AI faster. The issue is that what deepsea showed is that those advances won't last as long as we anticipated. So you may use more compute, for example, in the US to release that frontier model. But now instead of perhaps having a two year Runway on that China competitor, maybe it's only a year, maybe it's only six months. And so investing in compute with all of these resources to the extent we're seeing the US focus on, may be putting too many chips on one ingredient in this process with respect to talent. Scott's question, I think that we're realizing that having a robust workforce that knows about the ins and outs of AI and cultivating more AI expertise and is going to be a larger part of our conversation in the States right now, the sheer number of folks who have a PhD in AI or ML disproportionately end up in private companies and disproportionately end up in a very small number of private companies. If we're going to be leaders of AI moving forward then my own proposal coming to a Lawfare article near you soon is we need some sort of extension system like we had for agriculture back in 1940. The Smith lever act said, hey, we've got all this expertise, go help the community learn how to use these tools. And I think we need to be starting to think about these broader solutions to how we're going to cultivate AI talent not only over the next year, but what's our decades long approach to really making sure that more folks are literate in these tools and can contribute to not only thinking about advances with respect to LLMs, but other AI architectures? LLMs will not be the end of AI evolution and so we need folks who are thinking about what's the next AI architecture that's going to push us even further rather than just ending the conversation here.

[Scott R. Anderson]

Well, we are running short on time, but we would not be getting through our full agenda if we did not turn some attention to our third topic, and that is the question of foreign assistance. We saw the Trump administration impose an across the board, almost across the board, I should say. There are a few carve outs, 90 day pause on US foreign assistance just a few days ago. That was actually complemented earlier today by a somewhat comparable pause on a lot of other domestic disbursements, which we're about to see legal challenges on. I suspect that might be a topic for next week's episode, but we're not quite teed up for it this week. But at the foreign assistance side, we're now a few days into a freeze that is really across the board impacting all sorts of programs. There are carve outs, it doesn't affect direct food assistance, which is one category of foreign assistance United States provides. It doesn't impact various types of State Department derived security assistance to Egypt and Israel were carved out. And then it doesn't impact, perhaps most importantly, security assistance through the Defense Department. This is a Foreign Assistance Authority's kind of Title 22 restrictions, so State Department and USAID. But we have seen not only this ban be implemented, but we've seen pretty dramatic action taken against a number of USAID employees. I think as many as two or three dozen of them, maybe more, who were suspended essentially on suspicions that they were involved in somehow avoiding these limits or rebuting them. Whereas other people are saying no. They were simply working with their partners, the recipients of foreign assistance who implement the program's assistance funds, to try and overcome the challenges presented by the sort of sudden unannounced suspension which is notable for these group. And it's worth noting the United States, while it's not the percentage of its budget on foreign assistance is not as great as a number of other countries. It is kind of the number one source of general foreign assistance in the world. I think something like 40% of direct bilateral foreign assistance comes from the United States. So a major pause on it impacts a lot of sectors that rely on that foreign assistance, particularly in the developing world and other areas where foreign assistance plays a big part in various aspects of the local economy and provision of humanitarian services and other services to the public there. So let me start by turning to you on this, Peter, as kind of for a framing question. And I'll say I know this actually probably was not squarely in your portfolio when you were at the White House and at the State Department before that, but I'm sure you've thought about this and I'd be curious, kind of your general thoughts. Where does foreign assistance fit into the broad kind of foreign policy national security toolkit for the United States? Or how has the Biden administration, the Obama administration, you worked and thought about it it and how it fits in there. And what seems so dramatic about how the Trump administration is approaching it here because this is a kind of unprecedented action. It's reviewing these programs is not new, but suspending them is a pretty dramatic step we haven't seen prior administrations take. Talk to us about why.

[Peter Harrell]

Yeah. So let me first say I was at the State Department for six years, a number of years ago, one of the places you and I overlapped in.

[Scott R. Anderson]

Our for better or for worse careers.

[Peter Harrell]

And you know, I very much remember and I think it is sort of par for the course, as you say, Scott, for a new administration to say, hey, we're going to we're going to review everything that's being spent on foreign assistance and make sure it's aligned with our priorities. And, you know, maybe over time a couple of things get adjusted here. But I have I'm unfamiliar, you know, thinking back at least a number of decades with anything remote approaching the scale of this pause, which is, you note, Scott, covers essentially everything except a couple of categories, quite narrow categories of military assistance to the close allies. And I think that's going to have, you know, some real world ramifications if it doesn't get turned back on quickly for things like health programs, frankly, for things like, you know, security assistance, people guarding prisons that were helping support for crime kind of reasons in the Caribbean and in the Middle East. I think it will have real world ramifications if they can't sort this out and get at least most of it turned on quickly. When I think about what we do with our foreign assistance, I think of it as kind of falling into two large overlapping buckets. You know, the first is really about development assistance. You know, this is things that aren't particularly self interested from a US perspective other than we want the world to be a good place. You know, health assistance, USAID health assistance programs vaccinate and provide mosquito netting and things like that for millions and millions of people around the world. Right. So aids, for example, AIDS medication. So there's a lot of kind of development and health assistance and also kind of emergency refugee assistance. Then there is assistance that is kind of about building relationships with key allies. And some of this is development assistance, you know, helping our allies kind of grow economically and merge closer to us or be closer to us. And some of this is military and other security systems helping build the security relationship. So I think that this is going to have ramifications if they don't sort it out both for kind of global health and development goals and also frankly, if they can't sort it out for a bunch of US geopolitical goals with allies around the world.

[Scott R. Anderson]

Yeah. And I think part of this that's really important to bear in mind that people who haven't dealt directly with U.S. foreign assistance, and particularly in developing world or post conflict areas, may not be aware of or may not think of is what a delicate ecosystem it is in many ways. You have these foreign implementing partners, the United States writes the check, but then you have people who are trained to implement the programs, whether it's health programs, whether it's security, whether you have things like that. Doing so is very expensive. It's often very risky. And these organizations rarely if ever run on very substantial margins. Right. Because they're usually not being funded by private sector funds in any meaningful regard. There are some USAID implementers that are actually like for profit companies, but even then they don't run on deep margins, certainly on the USAID accounts because that's not something which US foreign assistance agencies want to see. You're not supposed to be stockpiling these funds for a rainy day fund. You're supposed to be using them to implement things. And when you pull a lot of this out, you are potentially really undermining the infrastructure you rely on to implement these programs. That's not always easy to rebuild. You think about what it takes to stand up in office in a post conflict area. Security costs, keeping security staff there Building logistics and supply routes. This is something I dealt with when I was in Iraq with the embassy and a couple of partners there. A variety of contexts is really costly, really difficult when you suddenly tear that infrastructure down because the spigot of funding ends, it gets way more expensive to stand back up. Really difficult. And there's a huge drag of implementing at. And that's why I think the operating assumption has always been presumptive continuity, because the costs of cutting this can be difficult. And that's what really seems to have been lost in this sort of decision. And in my mind, it strikes me as a classic decision that makes a lot of sense if what you think foreign assistance is, which a lot of its critics do, is just writing checks to needy populations. But it's not. It is funding complex programmatic activities that often are designed to fill needs the United States sees, and often, frankly, members of Congress of both parties see as being necessary and in within US Interest in one stripe or another. Eugenia, you pointed to one really interesting consequence of this that I thought really spoke to this point about how foreign assistance intersects with US Foreign policy goals. And that's around cyber issues, which we know has been an emerging issue really since the Obama days. Big focus during Trump and Biden days, really, and probably promises to be moving forward. Talk to us a little bit about what the implications have been in that space which you track so closely.

[Eugenia Lastory]

Sure. It's in my mind, particularly distressing because this is something that has been slowly gaining traction over the last four years. I think one of the things that the Biden administration has been able to do is really pay attention and invest empower the idea of cyber diplomacy. And so the State Department, through its Bureau of Cyberspace and Digital Policy, actually got its funding up to $19 million from $17 million. And they were actually doing something very interesting, very innovative. They were using that foreign assistance as a way to engage with other countries, support their cybersecurity goals from a perspective of we need to raise the floor if we all want to be secure, supporting that. I think one of the more common examples of this is the support that they gave to Costa Rica when they were facing a tough ransomware incident a couple of years ago. Recently they deployed a cyber incident response team to Costa Rica. And now all of that has faulted. We had Ambassador Fick, who was leading the bureau before the change in administration, describe this as a way to prove to partners the value of working with the United States to build consensus against some of the malign actors that carry out the attacks. This was supposed to be a way for the United States to engage with countries that were maybe engaging with China, that were getting some of the infrastructure done through them. And now all of that has stopped. And I think this definitely is a clear example of that falls in the category of what Peter was describing as those geopolitical category of foreign assistance. I think it's going to make it harder to re engage. It's going to make it harder for those countries to take seriously the values that the US Purports to represent when it comes to cybersecurity.

[Scott R. Anderson]

Kevin, that actually brings us back to a point you made, I think, in our first conversation of the segment about competition dynamics. Let me pull that observation into this context. What does this kind of pattern behavior have implications for, from your perspective about international competition dynamics and kind of the Chinese versus US narrative in particular?

[Kevin Frazier]

Yeah, I think Euania with respect to cyber also tees up the conversation with respect to just Internet access and things like support for undersea cable infrastructure and infrastructure that you're going to need for AI development and diffusion across these countries. And so who is the next call? If the US says the spigots turning off, who do you call next? China's already numbered to call. Right. And so thinking over these long term.

[Eugenia Lastory]

Ramifications, China was on hold, you know, like they are already there, right?

[Kevin Frazier]

Yeah, yeah, you don't even need to call. You can just wave to your friend down the street and say, hey, we're ready for your investment now. I do think this is going to be very unsettling as we think in particular about the fact that by 2050, and don't quote me on this, I'll just podcast about it. That's different. Right. I think almost a fourth or more than a fourth of the world's young population will be in Africa. And so if we're not actively investing in developing countries, for example, in Africa, and we see other countries like China actively stepping in there, that has huge geopolitical ramifications over the next decades as we see demographic changes accelerate in Africa and very much slow down in the rest of the world.

[Scott R. Anderson]

Yeah, I think that's a really good observation. I mean, we saw the Trump administration take certain steps like this its last time in office during the first Trump administration and it triggered bipartisan backlash. It took a little bit of time and in particular that was kind of at a later moment. We are in this honeymoon period where President Trump has more influence in his party. He doesn't need anything from his party yet because they're not actively advancing legislative agenda. The one thing he needs are nominees, and there's one or two that obviously might have some issues coming up in the near future, but they're not necessarily the ones he's most wedded to. He got Secretary Hegseth to through seemed to be a big concern. I do wonder. I just suspect that kind of pushback is going to come, is going to come quietly, but it takes time. And the question is 90 days is a long time to pause these indefinitely. Maybe they start making exceptions which they've left the door open to do. They get things moving. Hopefully they would as they clear programs quickly and say, oh, we do want to keep doing this, at least temporarily. Maybe we'll come back to it later. So we may see a correction on this sort of path if we see a bipartisan reaction, but we'll have to wait and see. It's still in a few days and there aren't any signs of a strong, strong pushback coming, at least from within Trump's own party that I've seen yet. Well, folks, that is all the time we have together this week, but this would not be rational security if we did not leave you with an object lesson or two to ponder over in the week to come until we are back in your podcatcher. Kevin, what do you have for us this week?

[Kevin Frazier]

Yeah, so I'll step back on my soapbox and put on my professorial credential and say I really just want to applaud programs like the Technology, Innovation, Law and Ethics program at Seattle U School of Law as well as AI and law programs at the University of Miami, the work that Mark Williams is doing at Vanderbilt, Case Western now requiring that one LS get a certificate in AI. All of these steps are incredibly important. If we want to have regulations that get the balance right between innovation, knowledge diffusion and ethics, we need smart people who actually understand this technology. And right now, a lot of law schools are just saying, hey, go experiment, or worse yet, don't touch that tool at all and cite to it like you have to cite to Google Search, as if that was something you would ever do. So we need to really accept the fact that AI is here to stay. And our knowledge institutions need to see that they have a key role in knowledge diffusion. And so I hope that more law schools look to Seattle, look to Miami, look to Case Western, Vanderbilt. You pick one and lean into this and help make sure that lawyers not only at the Harvards, not only at the Stanfords, know what AI is, know how it relates to the law and can contribute to these conversations. All right, I'll get off my soapbox now.

[Scott R. Anderson]

All right, well, from one soapbox, perhaps another soapbox. But maybe not. Hopefully not. Because Eugenia, you have of course been, throughout your time on the podcast, our video game correspondent. I know you tried to shed this title your last few appearances, but I'm hoping that you come with one last sign off for your last episode of Rational Security recommendation for the nerds out there who need to keep themselves entertained well into the night on their laptop. What do you have for us?

[Eugenia Lastory]

I'm back, Scott. I absolutely have.

[Scott R. Anderson]

Good. That's what I was hoping.

[Eugenia Lastory]

Yes, yes. This time around I am going to recommend an indie rpg. It's called Wildermyth and it is just this fantastic game. I know that I say this about almost all the games that I recommend here, but it's really cute. I think it has a very interesting design. It's a fantastic storytelling experience. And what I love is that as you build out your characters and yes, you do control several characters and you can shape them all to be whatever you want, you actually see them grow from their little naive self adventurer who wants to go out into the world and make a difference and grow into an incredible demigod that controls fire. And you will actually get to import them from story to story, which is really nice. You get like Easter eggs of your own previous campaigns. It is adorable. It doesn't take a long time. It was a very good palette cleanser for me after I finished all of my Dragon Age playthroughs before I started with my Mass Effect playthroughs. So, you know, if you're looking for something nice, you can just devote a couple hours to it and then be done. Wildermyth. I loved it.

[Scott R. Anderson]

Wonderful. A wonderful suggestion. You do not disappoint with your deep nerd DOM reservoir. We greatly appreciate it. But I will celebrate your last episode of Rational Security by trying to out nerd you with my recommendation.

[Eugenia Lastory]

We'll see.

[Scott R. Anderson]

Because I am going to recommend an indie role playing game, but not the type you play on computer games. I'm going to recommend the type you play with pens and papers. As some listeners know, I like like nerdy board games and different types of games. I never had time to play them. I have two kids. It just never happens. I'm honest, I'm a pretty busy guy. But I got one for my brother for Christmas that was so interesting. I actually just got a copy myself. I think it's worth recommending. It's got A crazy name. It's called Thousand Year Old Vampire. But it is essentially this journal that you read and gives you a bunch of prompts to write a story about your interaction about this vampire's life over like a thousand year period. And you draw these really crazy narratives. And it has this incredibly visually compelling guide because it kind of made to look like a 19th century volume journal with all these inserted pictures and things you incorporate into the narrative. It's absolutely fascinating. I haven't actually gotten to play it yet. I just flipped through the version I got my brothers, but I just got my own in the mail. How having broken down and ordered it myself, it's absolutely fascinating, interesting. And you play it entirely by yourself. So if you actually don't have a lot of time to play games like you do, but you still want a little escapist fantasy when real world gets a little bit too much, which occasionally I need these days, I think this is a promising avenue, so I'll throw it out there. It's hard to describe physically how cool it is when you look at it. It's really like a work of art. It's phenomenal. But online I think you can get little sample pages that give you a taste of it. I will out nerd you with that nerdy recommendation, Peter, I don't know how nerdy you're feeling. You can keep it professional and above board or get down into the morass with us. What do you have for your object lesson this week?

[Peter Harrell]

Yeah, I'm afraid I don't have any good game recommendations. You know, my kids are 7 and 12 and it's really the 7 year old who wants to play games with me. And they tend to be uno and you know, that level, that level of thing. But I was gonna recommend. I've been doing some research on, you know, different periods, how the US thought about industrial policy during different periods of its history and in different. To serve different objectives. And have just finished rereading a book I'd read a number of years ago but had sort of forgotten about called Freedom's Forge, which is a book about how in the run up to World War II, Roosevelt and a bunch of, you know, shipbuilders and home builders and things got organized on the. The industrial policy front to ramp up for war production. I think there are a bunch of interesting lessons as we think about the future and industrial policy in the US today.

[Scott R. Anderson]

Phenomenal. Phenomenal recommendation. Well, that brings us to the end of this week's episode. But Rational Security is of course a production of Lawfare, so be sure to Visit [e-mail] for our show page with links to past episodes for our written work and the written work of other law firms contributors, and for information on lawfair's other phenomenal podcast series. While you're at it, be sure to follow Lawfair on social media wherever you may socialize your media. Be sure to leave a rating or review wherever you might be listening and sign up to become a material supporter of lawfare on Patreon for an ad free version of this podcast and other special benefits. For more information, visit lawfairmedia.org support our audio engineer and producer this week was Kara Schillen of Goat Rodeo and our music, as always, was performed by Sophia Yan. We are once again the at Edited by the wonderful Jen Patcha. On behalf of my very special guests Kevin, Eugenia and Peter, I am Scott R. Andersen. We will talk to you next week. Until then, goodbye.

[Marc Maron]

Hey folks, it's Marc Maron from wtf. Today I want to talk to you about Boost Mobile offering reliable nationwide coverage backed by a 30 day money back guarantee. Love your service or get your money back, no questions asked. Boost Mobile offers the coverage, network speed and service you're used to, but at more affordable prices. Why pay more if you don't have to? You can get an unlimited plan for $25 a month that will never increase in price, ever. No price hikes, no multi line requirements, no stress. Visit your nearest Boost Mobile store or find them [e-mail] After 30 gigabytes, customers may experience slower speeds. Customers will pay $25 per month as long as they remain active on the Boost Unlimited plan.