Welcome back, good Tuesday morning. Thank you so much for joining us this morning. I am Maria Bartiromo. It is Tuesday, July 21, just before 8:00 on the East Coast. The Treasury Department announcing implementation of a new government-wide payment verification process to help stop federal payments being sent to deceased individuals. Nearly 5,000 payments worth approximately $99 million to dead people. Joining me now, the Treasury Secretary himself, U.S. Treasury Secretary Scott Bessent. Secretary, thanks so much for spending the time. Can you walk us through the Treasury implementing new safeguarding to stop payments to deceased people?
Good morning, good to be with you. Look, this is part of what President Trump calls his common sense agenda. It is common sense. President Trump issued an executive order for us to be able to go to the Do Not Pay database, to have interact with Treasury payment systems. As a result, so far we've saved $100 million – payments that did not go to deceased people that would have gone out. We think up to $350 million that we can stop before the end of this year. That goes along with the Vice President's task force on eliminating waste, fraud and abuse in government. GAO estimates might be up to $500 billion dollars, about 1.66% of GDP. That could go a long way paying down debt, providing more services, and this is just the start, Maria.
That is incredible. I mean for all of the fighting that we see back and forth between the Republicans and the Democrats about governing, allocating capital spending money on projects, to think all these millions of dollars, tens of millions of dollars that have actually been, you know, on the cutting room floor because of fraud and waste, that people just weren't paying attention to. That is all it is. People did not look and investigate the way President Trump directed his agencies to...
Look, in the Biden administration, HHS got rid of 50 or 60 people charged with monitoring fraud. What is important here is that we are stopping more from going out. So once money gets out, trying to retrieve it is very, very difficult. Stopping at the source is our goal. There are hundreds of billions of dollars here. This is the start. We are off to a great start. You are seeing the Vice President's task force, things all across the healthcare system, too.
For sure, we've got a spotlight on it, covering it because this is incredibly important, particularly as many in Congress complain about debt and deficits. Efforts to save, eliminate fraud, waste and abuse. Your take on the economic pressure you've been putting on Iran at Treasury. You said Treasury is fighting an economic war against Iran, sanctions against Iran. I want to get your take on that. You put major sanctions on Iran's shipping empire this month, sanctioning a network tied to Hussein, targeting more than 200 vessels in addition to the naval blockade. Tough stance, retaliation against Iran. Can you tell us what is left in terms of your purview? Sanctions, the economic warfare can still happen against Iran.
Very successful. Six weeks combined, eliminated Iranian Navy, the Air Force, a large part of their missile and drone stockpile, destroyed most if not all of their capability to replenish. We moved on to what we call economic one-two punch. Blocked. What we're doing at Treasury, tracking accounts all over the world. Froze a crypto wallet linked to IRGC the other day, about $130 million. We have found the moneyman for the Ayatollah, tracking the Ayatollah's properties around the world. We hope to be able to print his hundred-million-dollar-plus properties, show the addresses. We are preserving this for the American people. But most of all, we are constricting the regime. Their currency has collapsed to an all-time low versus the dollar; it is in free fall. And we think the inflation rate is upwards of 180% in Iran. So you know, the government is causing the people to suffer, and we are going to keep pressing, but also going to marshal resources, save the resources that we recover for the Iranian people when we get on the other side of this.
So how much of Iran's ability to finance the IRGC would you say has been disrupted? And are you prepared to put sanctions on, for example, Chinese banks, if you find out that Chinese banks, Chinese companies are continuing to purchase Iranian oil and trying to support the IRGC?
Well, we have sanctioned Chinese, what are called so-called teapot refineries, the private refineries. We've seen a substantial decrease in their purchases of Iranian oil. The Chinese in general, because of the prices, because they have a very large strategic petroleum reserve, have dropped their purchases of crude about 40%, directly pressuring the Iranian regime. And we put the blockade back. I tell you, the blockade is powerful financially, I think also very powerful psychologically. It cuts Iran from the rest of the world. We also have the ability to interdict Iranian dark fleet once out of the Pacific.
Incredible. We are watching all of that, Secretary. I want to move to AI, because of course we are producing programming special this week, talking to the most important companies and individuals in this AI race. And this amazing buildout. The Lindsey Group reported the Federal Reserve found the CapEx of Amazon, Google, Meta, Microsoft, Oracle reached $412 billion in 2025, or 1.3% of GDP. Some estimates for '26 north of $700 billion. One estimate $1.25 trillion in terms of spending on this buildout. Can you assess for us what impact of the AI buildout infrastructure that is happening massively, what is the impact on the economy?
Maria, this is a generational technological boost. President Trump came in and became one of the great supporters of this very important technology. And it starts with we have to have the energy for AI. President Trump is allowing the AI companies, the data centers, to build their own power generation. So we are seeing it there, we are seeing it up the supply chain. And this is kind of a once-in-a-generation buildout. It is very powerful. On the other side of this, we are going to see big productivity growth. And really, that is how you increase national wealth, is through productivity growth. Look, the U.S. is the home for this. And one of the things I wanted to say today is we have seen talk about open source models coming, threatening the large language models in the U.S. What I will tell you is this administration supports open source models, but what we do not support is IP theft. If we see, especially overseas models are stealing from our great companies, we have the ability to sanction them because of this theft.
That is a really important point. That is exactly where I was going next. The Wall Street Journal this morning writes top American AI executives are sounding the alarm on Chinese models. Silicon Valley in Washington is debating. The multibillion-dollar question: Should American companies be able to use Chinese artificial intelligence models, and what does that mean for IP theft, knowing we have been watching Chinese steal intellectual property from America and the West for decades?
We are finding watermarks of our U.S. large language models on many of the Chinese models, and that is unacceptable, Maria. So we are going to be looking at that in the coming days or weeks. But again, the U.S. government supports open source, but what we do not support is IP theft. The other question might be: Should companies that are using Chinese models have to disclose that to their customers?
Good point. The Chinese Kimi K-3 is competitive with U.S. models on some benchmarks. The U.S. is going to be holding AI talks with China in September. You are leading that delegation to discuss with Chinese counterparts. From Axios, quote, a secret Trump administration battle to fight Chinese AI, including behind-the-scenes moves by the Commerce Department, National Security Agency, National Cyber Director, saying even considering completing an executive order saying U.S. companies can only host Chinese models if they can guarantee security, containing liability if it were breached.
Again, you know, models have to be held to the same standard. U.S. models are more... Importantly, are they using stolen materials? There is a very technical AI word for it called distillation. We call it theft. They have hundreds of millions of hits against the large language models here in the U.S., and try to reconstruct their code and the thought process of the models. That is unacceptable.
The Chinese are coming out with open source models, and they are cheaper. For an ordinary company, not necessarily one of the big major companies with the wherewithal to spend whatever they can spend on their own AI infrastructure, they, you know, a smaller company may need to go cheaper, want to use the Chinese version. Are they doing basically what they did with, they dumped cheap steel into the market? Now with AI models.
