9/21/26: AI Nearly Starts China War, Diesel Prices Skyrocket, Ed Sheeran Mocked For Apology On Palestine

Breaking Points with Krystal and Saagar

The episode explores the dangerous intersection of AI hallucinations and military intelligence, detailing a near-miss conflict between the U

Key takeaways

• A U.S. military operation against a Chinese vessel was nearly launched based on an AI-generated hallucination regarding nuclear materials.
• The reliance on AI in intelligence can lead to 'packaging' errors where faulty raw data is polished into believable, official reports.

Main topics

• AI military risks
• Diesel price crisis

Notable quotes

The intelligence report... was, quote, entirely false, but it also, quote, almost started [a war].

Conclusion

The episode highlights a growing pattern of systemic instability cause

Transcript preview

Speaker 4 (0:00) This is an iHeart Podcast. Speaker 4 (0:02) Guaranteed human. Speaker 3 (0:04) Whether you're a seasoned NFL fan or new to the game, there's one place to keep up with all of it. The league's newest podcast, the NFL report hosted by me, Andrew Siciliano. It is your home for everything. Football breaking news, expert analysis, game picks, and hear from your favorite players to join me at an all-star cast of experts for everything you need to know from around the league. Get new episodes of the NFL report Monday through Friday, all season long. Listen to the NFL report podcast. on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts. Speaker 4 (0:35) I'm Mimi Brown, and every day on Front Page Podcasts, I help you make sense of the stories shaping our world. Each day, I break down the biggest headlines. Canada's trade fight with the U.S. just escalated again, and Americans could fill it in their wallets. And when a story deserves a deeper conversation, I'll sit down with experts, newsmakers, and thought leaders to help you better understand the issue behind the headlines. Listen to Front Page from the Black Effect Podcast Network on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts. Speaker 5 (1:07) Folks, there's important news, there's depressing news. And then there's the news we actually want to talk about. On our podcast, Bananas, we search far and wide for all the news stories that make you say, what, what? And then we tell you all about them. Doorbell prankster Speaker 6 (1:22) that tormented German resident apartments turns out to be a slug. Speaker 1 (1:26) Ding dong. New Speaker 6 (1:27) episodes drop Tuesdays with more than 400 episodes to catch up on. Listen to Bananas Speaker 5 (1:32) on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts. Speaker 2 (1:37) Hey guys, Ready or Not 2024 is here and we here at Breaking Points are already thinking of ways we can up our game for this critical election. We Speaker 7 (1:46) rely on our premium subs to expand coverage, upgrade the studio, add staff, give you guys the best independent coverage that is possible. If you like what we're all about, it just means the absolute world to have your support. But enough with that, let's get to the show. A Speaker 2 (2:02) stunning new report indicates that we almost started a war with China. Based on a faulty AI intelligence report, let's put this up on the screen from CNN. This is pretty stunning and underscores how while we have been talking a lot about the long-term risks of AI, which I think we should talk about and be concerned about, there are a lot of short-term risks involving AI hallucinations and human stupidity, the combination of which could be completely disastrous for the country and the world. So this is the headline here, exclusive. U.S. military had close call after using AI for false intelligence. report. The intelligence report circulated across the U.S. military this spring in the midst of the war with Iran immediately set off alarm bells. A Chinese ship in the Middle East was transporting components of a nuclear weapons program. The U.S. military swung into action with plans to intercept the vessel, according to four sources. According to two of those sources, armed members of the U.S. military were preparing to board the ship. Military planes were in the air. It was only just before the planned operation that officials dug deeper into the report, put together by a special operations command analyst, and found it had been generated with the help of artificial intelligence, and that a chatbot the analyst had used inaccurately identified the material the ship was carrying. The report, according to one of the sources, was, quote, entirely false, but it also, quote, almost started. A war. Any U.S. operation against a Chinese vessel could have risked spiraling into an armed conflict between the two nations. This is terrifying and completely insane. They had an AI chatbot that hallucinated intelligence about this ship carrying components for a nuclear weapons program, a Chinese ship. And we were this close to taking what would have been an extraordinary inflammatory. action risking war with China based on an AI hallucination. I mean, terrifying. Right. Speaker 7 (4:05) And the way that the chain of all this. So they say in this particular instance, the analysts queried a chatbot about some of the intelligence reporting. on the ship's manifest that had originated with the Special Operations Command Pacific, which was based in Hawaii. It's not even clear right now whether that chatbot was commercially available or a U.S. government one. I'm going to bet that it probably was one of those internal U.S. government products that ChatGPT, Claude, and Palantir have all built kind of inside for their use. The bot fused together. open source intelligence which secret signals intelligence in government holdings and reached its fateful conclusion about the material that the ship was carrying. The analysts then used AI again to package the findings into a standard intelligence report, the kind that is trusted by military officials. So you hallucinate it. I mean, this is classic, actually, why what's happening in colleges right now is so bad. What you do is you kind of take your raw research from ChatGPT or others. And then you have ChatGPT or Claude or any of these others. package the raw research into something even more polished. So it looks like all the back research has been done. But if the chain is faulty from the very beginning, then it gets created into the standard kind of briefing. The briefing then, look, these 60-year-old commanders are like, all right, they've been reading these intelligence reports for years. It looks exactly the same. They go, yep, intel solid. Let's launch the operation. Obviously, they know how that's going to work. Special Operations Command and the Pentagon then launch the entire mission. mission. And it's only until somebody who was like, hold on a second, they're sending, where's this coming from? That they start to look into it. The other problem is that. all of it was very decentralized, is it doesn't appear that any of this went very high up the chain of command. So maybe Hegseth knew about it. Did the White House know about it? I mean, to me, to my mind, if you're going to launch an operation on a Chinese ship, the president himself needs to sign off on something like that. I mean, that is such world-changing implications. But that's, again, this is the problem with getting into a war in the first place, is that's not feasible, by and large, for every single military operation you kind of delegate with. the bounds. And one of the bounds he probably set out was you don't even need commander in chief authority. If nuclear material is on its way to Iran, you go in, you go ahead and intercept it. Right. Which again is scary, but you could understand exactly how that Speaker 2 (6:25) seems that even that seems so wild. I mean, even if it's not China, any sort of, you know, nuclear material coming in from any, I mean, you would think that that would be. involvement all the way up to the highest level. And so, you know, the crazy thing is that according to a lot of research that's out there at this point, if this same intelligence analyst had gotten this intel from a human source, I bet you he would have questioned it more. OK, this is an extraordinary claim. How do you know? Where are you getting this from? But there is unfortunately a typical human bias in favor of the machines, where if the AI tells you, people are more likely to accept that without looking into it than if you get it from a human being. So we're seeing that, you know, with this near miss of an incident. And then the other place we can put D2 up on the screen, you know, obviously in the first day of the war, we a girls' school, mass murdering. By the way, those little girls would have been attending their first day of school in Iran literally today. And instead, they completely, I mean, just horror, horror all the way around. The worst attack on civilians in modern history from America. And the new reporting says that AI contributed to that. Now, you can't just say it's AI's fault here, though. This fault lies directly with the human beings. You had a rush to come up with an extensive target list. You have, as they say here, outdated imagery and an over-reliance on AI. So in their rush to spin up a large target list to launch this aggressive, offensive war of choice, they skipped over all of the steps that you would have wanted to take. to make sure that you're not mass murdering little children in their school uniforms while they're at school. And so because of that, you know, rush, the fact that also Hegseth has dismantled the office that is supposed to look into, you know, potential civilian deaths and make sure that there are safeguards there. And then because of this reliance on AI, you end up with an absolutely monstrous, horrific war crime. that was perpetrated on the very first day of this war. So you can see how there is already, where we are now with AI reliance, there is a combustible, tragic, destructive. combination of human stupidity, over-reliance, trusting these products blindly, and use of AI that's being fully integrated into government systems that in many ways has already been fully integrated into consequential government systems, including within the Department of War, where people's lives literally depend on it. Speaker 7 (9:11) It's really scary because I think that this is the most extreme example. But I was talking with a good friend of mine who was talking about this at a company level. And the amount of of AI that is now integrated into top CFO and CEO decisions, you just have to think about the consequences of this. Let's say that there was a potential AI bubble, but you're using your AI to power all your finances. Wouldn't the AI have a self-interest in telling you to keep funding it, even if it was gone? And I was like, wow, and he brought this up to me. I said, I didn't even think about that. But, you know, it's one of those where we've outsourced so much decision-making and so much research that what we think is an independent decision is not independent actually at all. It's only dependent really on the tools that you're using in front of you. Trump also, what's so become clear to me, and I was talking about this with several friends, like at the end of the day, why does Trump care? It's purely self-interested. The AI capex spend is the only thing floating this economy. That's it. He has to defend it. And the AI oligarchs, Jensen, you know, and all of these others, the safety stuff is just too inconvenient for the stock price. Like the fact that Sam Altman's not going public, that's not good for the U.S. equities market. You know, it's good for humanity, but it may not necessarily be good for the U.S. equities market. Anthropic, they need to go public and they need to keep all these data centers being pushed up everywhere. You can actually see this with Trump. Put D3. up here on the screen. So I won't even read all of it. You know, over the years, there've been many hoaxes, et cetera, blah, blah, blah, blah, blah. All right, now we'll finally get to AI. We will not in any way hinder or stifle the growth of this incredible industry. Rather, we will cherish it, help it, and watch it. as it grows. However, we'll also be looking for bad, and we can do that very easily with our existing criminal and civil justice system. For this purpose, I am forming the AI force, which, much like I did with Space Force, has been a tremendous success in my first term. To that end, I will be announcing in the near future an AI czar. Only high IQ individuals need apply. AI is the next industrial revolution or internet, but will be even larger and more impactful, possibly as much as 25 % of our country's GDP. we are leading China, the rest of the world, I intend to keep it that way. Ironically, people who are fully AI-pilled were saying, hey, 25 % is far too low, sir, to be pointing out. Remember, Anthropic said that its total addressable market is the entire US economy of $30 trillion. But that is why the positioning of the White House, and there's even been many reports, many people inside the White House are like, ah, I'm not so sure about this, causes real problems. in the middle of diesel and the gas crisis and inflation where it is right now, this is all we have. This is it. It really is. I mean, I'm looking right now. So gas is $4.50 a gallon. And let me see where the US equities markets are. Yeah, equities are actually up. you know, after market. Obviously, it's one day, etc. But, you know, we live in a world where the Baba Mandir Strait and the entire strategic situation looks so perilous for us and the U.S. consumer looks so screwed heading into a fourth quarter where retail usually is, you know, the backbone of the U.S. economy is people buying dumb Christmas gifts that they don't actually need. That's literally what keeps us all propped up. And our equities market is still up. in aftermarket and we could be heading for potential all-time highs nearly every other month with the S &P 500. It's only because of the tech stocks. Speaker 2 (12:41) Yeah, no, absolutely. The whole economy is a bet on this. And you look at the, we've talked before about this chart that shows as data center construction increases, all other construction decreases. I mean, it is eating the entire economy. And so Trump. who cares a lot about the stock market anyway, but has to care a lot about the stock market because the whole economic situation is quite dire already. That's why, of course, he's not going to put any limits on this. Of course, he's not going to rein the industry in. His entire presidency, the story of it, has been giving the tech oligarchs everything they possibly could want. And Jensen Huang, the CEO of NVIDIA, has really emerged as a villain here who's quite duplicitous, I think, in terms of the way that he the way that he presents himself and he has really gained Trump's trust and confidence. It's worth remembering, you know, NVIDIA is at the center of all of this circular financing. They want the shopping spree of companies coming to them, buying their chips to go on forever. That's what they want. So he has a direct financial interest in saying, oh, there's no problems here. We need to just keep going off to the races. The whole economy depends on this. Let's go, sir. And Trump is always looking for the person who's going to tell him. what he wants to hear. So Jensen seems to have really gained his trust and have his ear right now. And we played some last week of Trump calling him while Jensen was on stage at the all-in conference. And you could hear the way that he talked to Trump. Oh, Mr. President, don't worry. We'd never let that happen. You're doing a great job. You know, this is going to be amazing for the country. And that's the way that he is. In public, you can only imagine how slavishly, you know, worshipful he is in private to Trump as well. And meanwhile, just again, the total lack of seriousness of this man. Like, what can you even say? Put D4 up on the screen. He put this out. He wants us to vote on a new word for AI because he doesn't like artificial intelligence. He says that maybe we should rename it supreme intelligence or extreme intelligence. or a superior intelligence. Now, this was the second follow-up tweet where he says, well, supreme intelligence is losing really badly. People like superior or extreme intelligence. So we're taking supreme intelligence out of the running, deleting it as Speaker 6 (14:59) a Speaker 2 (14:59) qualifier, let you vote for the final two, superior intelligence or extreme intelligence. A fresh vote begins now. Oh, thank you, sir, for the opportunity to participate in the future of the country and have a democratic say over the direction we're going in terms of what we label this extremely powerful. powerful and dangerous technology being built by people who want to replace all of human labor and say that it may also destroy all of humanity. Thank you so much for that, sir. I appreciate it. See, Speaker 7 (15:25) to me, it's just clear that this is like Trump. He believes it's very similar to the Iran situation. Everything for him is it'll always work out and it's always a branding problem. Like, you know, it's not about the reality. It's not about the technology. It's like it's just about branding. Even in the way that he wants to go in the midterms, he genuinely believes, he believes in his core that he has done an incredible job and that it's all about his advisors not selling the war or his accomplishments properly. Like he thinks it's packaging. And you can understand why the person who became president through Earn Media and through The Apprentice would believe. something like that. He's Speaker 2 (16:02) also an old man who's not going to have to be here to deal with all of the fallout that the rest of us are going to have to deal with or that our children and grandchildren are going to have to deal with and clearly doesn't give a shit about anyone except himself. Right, Speaker 7 (16:16) and his family and who are also getting very, very rich. Filthy rich. Speaker 2 (16:19) You know, it is funny because they talk about the, you know, the AI technology as having like spiky intelligence is kind of the word for it where on the one hand it can solve a millennium math problem which is like, you know, the best humans were unable to. You win a million dollars if you're able to do one of these things. It can identify multiple zero-day exploits, which are like, you know, hackers can be paid hundreds to millions of dollars to identify these. And yet, on the other hand, you can ask it some basic thing and it'll just like completely invent an answer and lacks the intelligence to even think of like, hey, would the humans want me to hack into this company right now? There's some basic sort of morality and common sense intelligence that's lacking there. Trump is a little bit like that. His intelligence is very... spiky in some ways. Yeah. He's Speaker 7 (17:02) an Speaker 2 (17:02) idiot. The dumbest person on the planet. Like I have no doubt. Like he actually likes to use AI for all of us. Frickin boomer slop. Does he have any understanding of what this is actually doing, what it's capable of, of envisioning what the world may look like if the most dire, either positive or negative visions of this technology come to fruition? No. But he is this sort of like marketing branding. savant. And so it's part of what makes Trump very difficult for people to wrap their heads around, that someone could be so talented in some ways and so stupid at the same time. But he's kind of a mirror image reflection of the AI. Speaker 7 (17:39) Many of our presidents, actually, we'll show you one of them, who appears to be this, you know, great thinking article reader and all of this, the most well-spoken man you could ever envision. But then when he actually was president, couldn't literally do anything. But before I get to that, you know, you said something funny about the AI and all of that. I actually think that's the most human thing about AI and it reflects the people who built it. Because if you've ever met somebody who is top 1 % or top 0.5 % IQ, that describes them to a T. They're incredible at like mental math or they'll be incredible at like mental math, have a visionary ability to like look deeply. I think the most interesting thing about high IQ people is their like cross-domain knowledge and they can use things cross-domain to come to a conclusion which you literally never