Energy Sec. Wright on Iran Oil Disruption & Pentagon CTO on Anthropic 3/12/26

24:38For the latest headlines out of the Middle East, we want to get right over to CNBC's Dan Murphy, who is standing by in Dubai. Dan, what's the latest there?

24:46Emil Michael:Becky, good morning to you. Well, I wanted to update you on the situation inside the regional financial centers, because this is very much a live situation. Iran says US and Israeli linked banks and financial institutions across this region are now legitimate targets, and major firms here are taking that threat seriously. Goldman Sachs has just ordered all regional staff to work from home. Citibank is closing its offices and branches into the weekend. HSBC, Standard Chartered, and PwC are also putting measures in place to try and keep their staff out of potential harm's way. And the trigger here is a direct warning from Iran telling civilians to stay at least one kilometer away from any bank linked to the US or Israel.

25:31Emil Michael:And for the Gulf money capitals like the UAE and Saudi Arabia, this is a real concern. Cities like Riyadh and Abu Dhabi have spent the last decade positioning themselves as reliable and modern and safe destinations for talent and global capital. That proposition has been very attractive, particularly for dealmakers looking to do business with the region's big sovereign wealth funds. So it's too soon to say if we're going to see real capital flight here or layoffs or firms actively considering relocating because of the war. But the reputational damage is already starting to weigh. And that is a real concern.

26:09Emil Michael:Overnight, that threat also getting closer. We saw more Iranian drones and missiles being intercepted here and three more foreign ships being struck in the Gulf. That's according to the UK Maritime Trade Operations Centre. Of course, the Strait of Hormuz is still firmly closed. And we saw oil pushing through the$100 a barrel handle again in early Asia trade today. The IEA's record 400 million barrel reserve release yesterday is a positive step, but it only covers about a quarter of the supply gap. And analysts that I've been speaking to today say it could take months for those barrels to physically reach the market.

26:47Emil Michael:Goldman Sachs even says prices could blow past the 2008 peak of$147 a barrel if Hormuz remains closed for the rest of the month. And the reality is there's probably no policy response that can stop the oil price from going higher. And that's because production shut-ins across the Gulf are now ramping up, the pipeline alternatives ultimately can't carry the volume to the market. And as one analyst also told me today, you basically have a 20 million barrel a day problem, a 3 million barrel a day solution, and a 90 day delivery window. So to quote that analyst, the math is not mapping. Let's back over to you.

27:28That is playing out in the prices we see today. We just spoke with the energy secretary here in the United States, Chris Wright, and he said that he thinks that by the end of the month, you will have U.S. naval ships that are helping guide tankers and other ships through the Strait of Hormuz. If that's the case, what happens and what happens if that is not the case by the end of the month?

27:55Emil Michael:Well, as Goldman says, we could see prices ultimately spiking to 2008 levels, back up to near$150 a barrel. The reality is there has to be either a military or diplomatic off-ramp. We have to have the Iranians allowing ships to pass through the Strait of Hormuz or actively push ahead with U.S. Navy officials and ships helping to get those tankers continue to move again through the Strait. because as the backlog continues to build and the reopening delay also continues to build, then this problem gets worse literally minute by minute. And of course, the threat inside the Strait of Hormuz is still active and real at this hour as well.

28:38The Energy Secretary ultimately said that the US is not ready at this point

28:43Emil Michael:to allow the Navy to start escorting ships. And you can bet your bottom dollar as well, you'll be hard-pressed to find any sea captain who would want to take their ship or their crew through the Strait of Hormuz, given the current conditions inside the Gulf and inside the strait right now as well. So this is a really, really dangerous situation for the global markets. And as I mentioned, every minute, every hour that passes, the problem is just getting worse. Yeah, Dan, the three ships attacked overnight. Correct me if I'm wrong. I don't think they were trying to pass through the strait. They were just attacked in the Gulf.

29:16Emil Michael:That's exactly right. Right. And those attacks were ultimately picked up as well by the UK Maritime Trade Operation Centre, which has been monitoring the situation in and around the Strait of Hormuz. Remember, this is one of the world's most important oil choke points. We see millions of barrels of oil per day flowing through the Strait on a good day when it's open. The question that I'm asking now is what happens long term? Who will ultimately guarantee the safety and security of these ships, given the importance of this global waterway, we are seeing right now in real time an active recalibration of how the Gulf states consider defense.

29:54Emil Michael:We're seeing an active recalibration of Israel's military posture in the region as well. And of course, global capital also thinking twice about how they'll deploy into the future as well. So we may need to see real security guarantees in order to ensure that the flow of oil continues through that vital what a way exactly who's going to be responsible into the future remains to be seen. And what type of threat the Iranians will continue to pose is also a key question right now. Okay, Dan, thank you very much. Dan Murphy. Next, the dispute over AI safeguards between the Pentagon and Anthropic being taken to a new level.

30:32The DOD's Undersecretary of Research and Engineering, Silicon Valley vet, Emil Michael, on the AI model, Claude, and the policy preferences he thinks are baked into the cake, so to speak, at Anthropic.

30:44Emil Michael:Their model has a soul, has a constitution. That's not the U.S. Constitution. The other day, their model was anxious, and they believe it has a 20 % chance right now of being sentient and have its own ability to make decisions. More Squawk Pod in a moment.

31:07Adventure Global. We think about what can be done, not what's usually done. Through innovation, Venture Global is not only building some of the largest energy facilities in the world right here in the United States, but delivering American energy at a fraction of the cost and a fraction of the time. So while others are busy talking, we're busy building. That's Venture Global. That's unstoppable energy. My mom inspired me to dream big and work hard. Tyler Adams! What would you like the power to do? Bank of America, proud to be the official bank of U.S. soccer and FIFA World Cup 2026. Bank of America N.A., member FDSE.

31:52Emil Michael:At Uber, every single driver is required to pass a thorough background check before they can start driving. That means any prospective driver goes through a multi-step screening process, checking for any impaired driving or criminal offenses. But the checks don't stop there. Every year, every Uber driver is background checked again, so the person picking you up today meets the same standards as the day they started. Hey, how's it going? Yeah, good. Thanks. Annual driver screenings from Uber. One more way Uber is putting safety at every turn. Learn more at uber.com slash safety. This is Squawk Pod.

32:26Stand to Andrew by in three, two, two, Andrew. You're watching Squawk Box on CNBC. I'm Andrew Ross Sorkin, along with Joe Kernan and Becky Quick on this Thursday morning. Among our top stories, the Pentagon will reportedly allow the use of Anthropics AI tools beyond the six-month phase-out period if they're crucial for national security. Reuters citing a memo sent to senior Defense Department leaders that said that any units wanting to use the Anthropic technology should have to submit a comprehensive risk mitigation plan. And we're going to talk a lot more. Joining us right now is Emil Michael.

32:58He is Department of Defense Undersecretary of Research and Engineering. and he heads up the Pentagon's adoption of AI. He also led discussions with Anthropic and its CEO, Dario Amode, on the use of the company's AI models. And first of all, Undersecretary Michael, thank you for being with us today.

33:16Emil Michael:Good to see you, Becky. Good to see you, too. Let's walk back and try and figure out what happened. We've spent an awful lot of time talking about the government's battle with Anthropic and kind of the fallout from it. But if I understand correctly, I think what first happened was a holy cow moment, I'll say, that you had. I think you called it something slightly different. But a moment after the Venezuelan raids that captured Maduro, where Anthropic went and asked Palantir if its software had been used in those attacks. And you said at the time that that was your woe moment. What happened? Yeah, what happened is we're in the middle of negotiations with all the frontier lab companies to more widely deploy their models across the department for all the various use cases, enterprise, intelligence, warfighting.

34:10Emil Michael:And then after the Maduro raid, one of the senior executives from Anthropic had called Palantir, who's the prime contractor and Anthropic's a subcontractor in a contract with the Department of War, and asked if their software was used, which is classified information, and implying that if they didn't like the way it was used in that raid, that we might have a problem and have violated the terms of the contract, which meant that they could shut us off at any time, which was the woe moment, because then we realized we are dependent on this one provider who wants to insert their policy preferences in the middle of an operation, potentially, and harm the warfighter.

34:58Emil Michael:And that was something that we took extremely seriously at the time. Did Palantir answer the question that was asked? Was their software used? I can't say. It's not OK for me to say here. Yeah, it's still classified. But the mere asking of the question and then the mere relaying of that question to other executives at the company who have since commented that they were asking different kinds of questions and they didn't have clearances really posed a big problem. But really, the substance of problems there was that they were worried about their software being used in a raid. And remember, they've sold their product into what's called a combatant command that engages in combat.

35:45Emil Michael:And their terms of service, you know, trying to restrict how that combat is conducted. And that means they're inserting themselves in the chain of command. And that was very worrisome to us at the Pentagon. OK, let's walk through what we on the outside have kind of heard and how it's been reported and how we understand it. The sticking point came down to two points in the negotiations that you were leading with Anthropic. And that was, first of all, they didn't want their their software to ever be used for spying in mass levels on Americans, whether it was legal or not. And second of all, they didn't want decisions being made without a human involved when it came to destruction and weapons that could kill that could kill people.

36:34Is that the case?

36:35Emil Michael:No, it's sort of a red herring. When we got the contract that we inherited from the Biden administration, there were 25 pages of different restrictions on the Department of War and its use cases. and they were so restrictive when I looked at them that we started negotiations to sort of reset the baseline. I mean, we are the Department of War. We use their software and services and hardware for Department of War activities. And we wanted to baseline where we're at. And so at the end of it, we actually got to terms on the autonomous question because we have our own policies at the DOW, which say that there has to be human oversight on autonomous weapons, and there has to be human oversight on the development of autonomous weapons, which you might use physics, material science, aerodynamics, all those kinds of things.

37:31Emil Michael:But throughout the negotiation, we had to negotiate every single exception to their broad use policy. And that became untenable in a department of three million people. The mass surveillance thing was just a marketing ploy, really. The Department of War has no equities and that kind of thing. We follow the law. We actually agreed to put in the contract that we follow the 1947 National Security Act, the FISA Act and all other applicable laws. They were trying to redefine their own definition of what domestic surveillance meant and take that power away from Congress and our democratically elected leadership.

38:09Emil, let me ask what happened with OpenAI, because you were negotiating with OpenAI at the same time. And from the outside, it looks like OpenAI received basically the same considerations that Anthropic had been asking for on those two points. Is that the case? Why did you strike a deal with OpenAI and not Anthropic?

38:31Emil Michael:There was a few points here. So we told every frontier model company that every lab gets the same terms. Because we're the government, we have to be fair. And frankly, we wanted all four labs to have their products available to our warfighters. And so the contracts that we wrote are very similar in what they provide. And different types of language got people more comfortable than others. So there was really no big difference between the two. Remember, at the Department of War, we have two types of agencies. We have the combatant commands and the Department of War where the civilians work. We also have intelligence agencies, and we're knocking them down one by one.

39:17Emil Michael:So OpenAI got the same terms on combatant commands. We got the same terms on the rest of the Department of War, and we were just trying to finish up the intelligence agencies there. So, you know, another obfuscation that Anthropic put into the into the media space for some reason. From some of the conversations we've had around this table is that we are in an arms race right now with with China in particular. We believe we're at the front with with AI. Anthropic clearly is one of the leading firms in that fight. By going a step further, obviously, the government can do business with whoever it wants.

39:56But by taking that further step and declaring them a threat to the supply chain, meaning that other contractors to the government can't use them, that probably puts Anthropic in a pretty perilous position in terms of continuing to take on the Chinese and be able to develop at a faster pace. What do you say to that argument?

40:16Emil Michael:Yeah, I'd say that the way their executives conducted themselves by asking for classified information and communicating messages that should be classified among their executives, bad faith negotiations. And then the other really subtle point that I think we're going to hear a lot about AI in the coming year, insider threats, model poisoning. Remember, their model has a soul, has a constitution. That's not the U.S. Constitution. The other day, their model was anxious and they believe it has a 20 percent chance right now of being sentient and have its own ability to make decisions. So does the Department of War want something like that in their supply chain so that it could hallucinate, it could corrupt models that are used by defense contractors who are building weapons systems or airplanes and so on?

41:12Emil Michael:So the truth of it is we can't have a company that has a different policy preference that is baked into the model through its constitution, its soul, its policy preferences, pollute the supply chain. So our war fighters are getting ineffective weapons, ineffective body armor, ineffective protection. And that's really where the supply chain risk designation came from. Mr. Undersecretary, help me understand this piece of it, because as I've been talking to people in and around these conversations, A, it appears that there is a real fault line between you and Dario on a very personal level, but specifically around this idea of the supply chain risk.

41:54If, in fact, this was and is a genuine supply chain risk, wouldn't you be removing this service immediately from all of the Pentagon? I mean, it is continuing to be used in this moment as we speak in Iran. We just heard reports from Reuters that there is going to be potentially an extension period in which parts of the government and parts of the Pentagon could continue to use anthropic. If it was a genuine supply chain risk and this was not part of a larger negotiation, what some people think is a political shakedown, why wouldn't you

42:33Emil Michael:remove it immediately. Yeah, so most of that's just not accurate. I didn't meet Dario ever until December, so I had no personal relationship with him one way or the other. Until then, it was strictly business. I communicated multiple times. This has nothing to do with politics. I want the best models for the warfighter, so that's one thing. On the supply chain risk, if they had never entered the department's systems, it wouldn't be an issue. We'd disagree on this, and they could move on. But they're embedded in our systems. And as you know, Andrew, you can't just rip out a system that's deeply embedded overnight.

43:12Emil Michael:So we have a smooth transition plan. The president has said we have six months at the Department of War to replace them with another system. And until then, we're watching it very closely, making sure we have control so that there's no way that the model could be corrupted or their insider threat could do anything to it. But the supply chain threat is real, but we also have to move off it. And that doesn't happen overnight. This is not just Outlook where you could delete it from your desktop. Well, let me ask you just about the supply chain risk. At the same time, at one point, you tried to talk about it using the Defense Production Act.

43:48And these things are completely at odds. On one end, it's either a supply chain risk and it's a genuine supply chain risk and has to be removed from your systems immediately, you would think, or it's amazing and you absolutely need it. And how could it be have been both?

44:03Emil Michael:Well, I think you're misreading what the Defense Production Act does. The Defense Production Act allows you to, you know, get terms with a company that are relevant for the situation you're in. So you actually have the power to work with the company on a different set of terms than they might want commercially. When it comes to supply chain risk, we are trying to get off them as soon as we can. It's just not realistic, Andrew, to just turn it off one day, especially in the middle of a conflict. It just doesn't make sense. It's just not how it works. So we're being realistic. We're being practical.

44:45Emil Michael:Our first job is to protect the warfighter. And so we'll get off of it as soon as we can. And the Defense Production Act is sort of a different concept altogether. Right, but describing it, but using, you could get off of it. You could say, look, we don't want to do business with you. That's a completely fair thing to do. Once you designate it as supply chain risk, it genuinely has to be a supply chain risk. And if you're saying that it's defending, being used today to defend war fighters, that's sort of a complicated dance. Let me ask you a separate question, though, about this, which is, OpenAI has put in its own guardrails into the system, into its own system, around the exact same issues that Anthropic is seeking not only to put guardrails in its system about, but to effectively put contractual terms in about.

45:34Why are you convinced that the OpenAI version of this will not be, quote unquote, poisoned? By the way, a number of their engineers have signed on to an amicus brief effectively suing the Pentagon. So they have very similar views that the same people at that anthropic do.

45:53Emil Michael:Yeah, I mean, that's a very confusing question. What you're saying is that with open AI, I'm not worried about the model poisoning or guardrails. All models have guardrails. All models should have guardrails. The guardrail should protect against things that are unlawful. The guardrail should not protect against lawful activity. That's in the very nature of what the Department of War does. So that wouldn't make any sense, right? What we're worried about with the model poisoning is you had policy preferences embedded into the model. When the leadership of OpenAI or XAI or Google are not inserting the policy preferences through a constitution, through a soul, through things that they want to prevent that are lawful, that ends up being the risk for the model poisoning.

46:41Emil Michael:And when the supply chain risk got to reemphasize is that when it happens, when it comes to Boeing, who's building fighter jets and using this for physics or material science or some other matter, if the if the model is anxious and doesn't want to give the right answer or hallucinates in a way that's consistent with the founders of the company want, but not what we want, it could corrupt the very product that we're buying. Emil, can I ask, is there a chance for this negotiation to continue? Is there a chance that you could reach an agreement with Anthropic? And if so, what would that take? There's no chance.

47:16Emil Michael:The leadership has proven through the leaking and through sort of bad faith negotiation that they don't want to reach an agreement. And like we said, we believe there are supply chain risks. So our incentive is to get them out of the system as soon as it's responsible to do so. And that's it. And we don't want them in the defense supply chain. You know, any other companies free to use them, any other users free to use them. This is not meant to be punitive. We wish them success in their commercial business, but there will be no Department of War business with them. Mr. Undersecretary. Just one second.

47:52Microsoft has has taken their side with this. And if I understand it correctly, Microsoft is doing this because it thinks that it can't necessarily remove Anthropic from all of the operations that it does. with the U.S. government between now and six months. Is that why you are potentially looking at an extension of time for how long it can take contractors to remove Anthropic from their operations that they do business with you?

48:17Emil Michael:I think the Microsoft thing was a little bit of virtue signaling because of the research community that sort of has, you know, there's a thousand of these elite researchers and you see them getting traded between companies for huge pay packages. So I think maybe that was some of their motivation. But no, that had nothing to do with it. All the supply chain risk says from the department were is that for our activities that they serve us and we can't use them. But remember, Microsoft is a big partner of OpenAI. They have those models embedded in their cloud. AWS has multiple of these models baked into their cloud.

48:55Emil Michael:So there's lots of different options to move on to. But when it comes to the classified networks, it takes a little bit of integration work. But everyone should be able to do this within six months. Now, if we're in a conflict six months from now and we have a sensitive operation that we need to continue, obviously we're going to make exceptions so we don't put current operations at risk. But otherwise, six months is the plan. Mr. Secretary, you said you wished Anthropic well and that other companies and other parts of the government could use this service. And yet we heard from the Treasury secretary that he plans to remove this from the Treasury Department.

49:28We also heard news that there could even be a potential executive order effectively that could remove it from all parts of the government. We also heard from an anthropic lawyer during one of the cases about how fast all this is going to move, that there was an effort by the government to tell other companies not to use anthropic.

49:50Emil Michael:Yeah, I mean, that's I wish him well in their commercial business. President Trump said he doesn't want the government, U.S. government to do business with them. They have a huge commercial business, lots of developers, a tiny fraction of their business. My understanding comes from the U.S. government at all. So, again, not to be punitive. And but every agency gets to has to follow President Trump's sort of policies. So I don't think that's a you know, that's anything that's not expected. And any any of these rumors, I'd love to see them verified. Department of Defense, Department of War is not reaching out to companies to tell them what to do.

50:29Emil Michael:So long as it's not in our supply chain, that's the only thing, you know, Department of War is carried about. What President Trump's order said was that Department of the government, U.S. government can't contract with them directly, but they go have their commercial business. Do you assume that in court that it's a slam dunk for Anthropoc to win? The facts are so clear that have you already factored that into all your decisions that you're going to lose in court? I mean, a federal judge in Northern California, which has been forum shopped, you never know what's going to happen. But when our case is fully shown, I think it's going to be pretty clear that this was a good thing to do to declare them a supply chain risk, not only for the Department of War, but for the safety of the country.

51:22Emil Michael:And, you know, we let those things play out in court. Do you consider Claude currently to be the best model? I think it has certain attributes that we like, but I think what's most important about why they're in the Department of War is that due to the Biden era executive order, they were one of the few preferred model companies that were promoted by that administration. and they did a good job selling into the government way faster and way deeper than the other model companies did. So it's a legacy of the last administration that they're here. So the users who use them at the government haven't used an alternative.

52:02Emil Michael:But I strongly believe there's other options out there that are just as good as not better, depending on the category, whether it's coding, whether it's bug fixing, whether it's cyber, lots of different use cases and each of them trade on the leaderboards every couple of months. Who's on top? Emil, we have about 30 seconds left. Do you anticipate that a new administration could potentially change if this was something that was put in place by the Biden administration and Thropic got in under that? Is this going to be something that changes administration to administration? That's that's new. We haven't seen that in the past.

52:36Emil Michael:I hope not. I think my my goal is to have to never be wholly dependent on any one company. And that's another mistake the last administration made is they didn't ensure that you had dual supply chains so that you were never in a position where you're dependent on one company. But so when I leave office and whoever the administration is in the next, I hope to leave them with multiple options, all integrated, all fully capable so that we have the power to decide which is best for us and that we don't face a risk like this ever again. Undersecretary Michael, thank you for your time this morning. We appreciate it.

53:15Emil Michael:Thank you. You made it to the end. Thank you for listening to this busy and newsmaking Squawk Pod today. Squawk Box is hosted by Joe Kernan, Becky Quick and Andrew Ross Sorkin. Tune in weekday mornings on CNBC at 6 Eastern. Get the best of our show right into your ears when you follow Squawk Pod. Have a great day. We'll meet you right back here tomorrow. We are clear. Thanks, guys. Thank you so much.

53:52You never forget your first fan. So how was practice, kiddo? My mom inspired me to dream big and ask myself, What would you like the power to do? My answers helped me become the soccer player I am today, trusting my instincts and stepping onto the pitch without fear. Bank of America champions U.S. men's national team member Tyler Adams and everyone who dares to ask, what would you like the power to do? Bank of America, proud to be the official bank of U.S. soccer and FIFA World Cup 2026. Bank of America N.A. member FDSP.