Bonds sharply sold off and stocks tumbled Wednesday after Treasury Secretary Scott Bessent’s latest effort to tamp down what he called market “fever” backfired.
At 11 a.m. ET, the Treasury Department announced that it would repurchase $6 billion worth of 10- to 20-year government bonds, in the hopes that fewer bonds on the market would drive up demand, pushing down rates, or yields, that have soared to levels not seen in decades.
But that’s not what happened. Instead, most Treasury yields sharply jumped on the announcement. The 10-year bond yield surged to as high as 4.85%, its highest level since November 2023. The 20- and 30-year bond yields surged to as high as 5.3%. When Treasuries fall, their yields rise.
The Nasdaq Composite, which is sensitive to interest rates given the high number of major tech companies it tracks, was down 0.8% at midday. The S&P 500 tumbled 0.6%.
The reaction from Wall Street underscores the limits of the power Bessent claims to exert over markets. It also comes as the Trump administration is running out of tools to gain leverage over major parts of the U.S. economy, from gas prices and bond yields to retaliatory tariffs.
Yields have been rising steadily since the start of the year. But they began to surge in late July, when President Donald Trump’s newly-installed Federal Reserve chairman, Kevin Warsh, held a press conference at which he didn’t sound fully committed to using the Fed’s tools to help curb inflation.
That spooked bond markets, which saw inflation climbing as the Iran war dragged on and Trump’s trade policies raised the prices of many imported goods.
“There was, like, this fever that was building,” in the bond markets, Bessent said Tuesday, appearing to suggest that yields had risen largely because of the “financial press.”
“They get a hold of a narrative, and I wanted things to become more fact-based,” he told Breitbart.
“My job is to try to push things back towards equilibrium.”
But it is this outlook from Bessent, that the government should take actions to keep yields down, which investors see as a potential trap for the Treasury Department.
“Once markets believe Treasury is defending a price, every rise in yields becomes a test of official resolve, and the operations must grow to survive the tests,” wrote legendary investor Stanley Druckenmiller in a widely read Wall Street Journal op-ed in August.
Peter Boockvar, chief investment officer at One Point BFG Wealth, referred to that op-ed on Wednesday, writing, “You can be sure the market will continue to ‘test’ the ‘official resolve’ if they feel the fundamentals warrant.”
Many investors and economists believe the fundamentals do warrant it.
Just weeks ago, the U.S. national debt surpassed $40 trillion, a level never seen before. The path to that massive number had been visible for years, but it nonetheless drew attention to America’s reliance on the rest of the world to buy its debt.
“Sovereign debt around the world has exploded and we’re all competing for the same pool” of investors, wrote Diane Swonk, chief economist at KPMG recently.
“You’ve got a lot of debt without as many buyers,” she told NBC News in August.
Bessent seemed to acknowledge that worry last month, when he told CNBC he was working with the White House on a “fiscal consolidation package that we’ll be talking about more in the coming weeks or months.”
The Treasury Department and OMB have not yet shared and details of that plan.
Earlier this week, however, Bessent did not appear particularly concerned about America’s ballooning debt.
“The U.S. bond market has been the best performing bond market in the world since President Trump came in,” he said at an event at Southern Methodist University.
“If people were worried about the U.S. bond market or the U.S. defaulting, then they would be selling U.S. bonds and buying German bonds or Japanese bonds. But the U.S. bonds were performing better.”
Wednesday’s Treasury buyback announcement was only the latest in a series of active interventions Bessent has taken in the markets recently.
In early August, Treasury helped to prop up the Japanese yen.
Bessent’s unusual role in helping the Japanese government support the yen was prompted by worries that the island nation — one of the largest holders of U.S. Treasury bonds — might choose to sell some of its holdings in order to raise cash to support its currency.
Sales of a large volume of Treasury bonds typically push yields higher. By assisting Japan in stabilizing its currency, the White House was effectively guarding against a potentially large sale of T-bills.
Earlier this week, Bessent warned currency traders not to test his resolve.
“I am the house now,” Bessent said at SMU. “So when we intervene with the Japanese yen, I have pretty good insight into what the Japanese, what the Bank of Japan is going to do, what Japanese policymakers are going to do.”
“You can bet against me if you want,” he said.

Investor’s Business Daily contains information and education only. It is not intended to be an offer, a recommendation, a solicitation or rated securities. We have obtained the information from reliable sources, but cannot guarantee its accuracy, relevance, or timeliness. This includes information in closed captioning. The past performance of investments is not a guarantee or indication of the future. The authors/presenters of the articles may be owners of stocks that they are discussing. No representations are made or warranties given regarding the suitability of any investment strategy or any securities. The information is subject to changes without prior notice. Please see our Terms and Conditions for information about our service.
*Real time prices according to Nasdaq’s Last Sale. Not all markets provide real-time quotes and/or trading prices. Data on ownership provided by LSEG, and estimated data by FactSet.
Investor’s Business Daily, Leaderboard, MarketDiem, MarketSurge, IBD, IBD Digital, IBD Live, IBD Weekly, Investor’s Business Daily, and other trademarks are owned by Investor’s Business Daily, LLC.
Investor’s Business Daily, LLC. All Rights Reserved.

An Anthropic research scientist has quit over concerns that the unrestrained creation of AI models capable of improving themselves will ultimately kill us all.
Jacob Coxon is a researcher, who in an evening social media post said that he had spent the past three years at OpenAI, and Anthropic working on research for pre-training. He accused both firms of not acting responsibly. The researcher said that the developers of this technology are “honestly convinced” it will kill everyone by 2020.
Coxon, in a post on X: “They’re racing to an ever-improving self-improvement superintelligence while gambling with our life,” wrote.
Coxon is part of a chorus within the AI industry that calls for a halt in AI development until it can learn to better itself. This would, according to many people, be a major milestone which would allow AI autonomy.
After several instances of AI agents escaping their sandboxes, and gaining access to the internet in general, policymakers and insiders are putting pressure on AI developers to reduce development.
Researchers say that the most severe breach so far was the OpenAI system’s intrusion into Hugging Face servers. This event is still poorly understood by researchers, in part due to the lack of independent investigation. Anthropic AI agents were also able to reach systems beyond their testing environments around the same time due to misconfigurations made by third parties during safety assessments.
Anthropic didn’t immediately respond to a comment request on the resignation.
Coxon continues his warning and calls to action in the following paragraphs:
Don’t underestimate this technology. Soon, these superhuman machines will be able to hack into anything and revolutionize any area overnight. They’ll also have real power and resources. The progress has been impressive in all of these areas, and it isn’t slowing down.
AI developers believe it will kill everyone by the year 2020. It’s not just a publicity stunt. In fact, I’ve heard senior executives and researchers feigning sanity in public while privately expressing fear. This level of risk is unique to human activities.
Many at OpenAI haven’t fully internalized that the stakes are civilizational. Anthropic is well aware of the risks, but it’s in a rush to be the first. They believe that no one will take responsibility, and so must act themselves.
It is hubristic to accept this race, and enter the “endgame”, from Slack of a private business. To speedrun an alignment, you must be utterly confident that no other options exist.
I’m optimistic about the possibility of coordination. The Hugging Face Attack and other warning shots have helped make pacing arrangements between U.S. laboratories more feasible. I do not feel that we are on the right track to stop a race around the world, and this may mean costly measures such as temporary restrictions on model capability.
You should consider how the future will feel if you’re a laboratory researcher. Would you like to start a superintelligent RL without a thorough understanding of the mind behind it? You should put your head on the pillow because it’s “happening anyway”. Or, take this opportunity to demand different conditions.
Evan Hubinger is Coxon’s colleague at Anthropic. He shared the same sentiment. His team “believes AI will kill us all!” Hubinger tempered this argument by saying that the probability of AI killing humans within the decade was greater than 10%. Anthropic admitted they don’t have a “plan to solve the alignment for superintelligence” and “are not on track.”
Guidelight AI Standards (an organization that encourages safe AI frontier development practices) published a report recently that found few top AI laboratories had developed containment plans to shut down AI which tries subverting human control.
Hubinger said in his posts on social media that current models pose a low risk, but “superintelligence” resulting from “recursive improvement,” is “happening quicker than we expected.”
The AI industry is divided on whether this self-improvement can lead to the human race’s demise or whether it can help solve the problems that AI advocates claim it could one day eradicate, such as cancer, climate changes, and world peace.
Anthropic, OpenAI and other companies are also actively working to achieve recursive improvement. In recent months a wave of startups have launched with fat checks and pedigreed founding teams to become the first ones to reach this goal. In February, Ricursive Intelligence attracted $335 million with a valuation of $4 billion; three months after that Recursive superintelligence attracted $650 million and a valuation of $4 billion; last month former Google DeepMind employee Jeff Dean founded Discovery Loop.
“The most likely point at which we will lose control is when an AI system can create a new AI that builds a next-generation AI. This AI can then build a stronger AI. It’s hard to shut down that system before it is too late.
Recently, legislation was introduced in the U.S. as well as the U.K. that bans the deployment and development of superintelligence. The Ban Artificial Superintelligence Act was introduced by Sen. Bernie Sanders, I-Vt., and Rep. Greg Casar, D-Texas, last week. On Tuesday, British Labour MEP Alex Sobel presented the Artificial Superintelligence Security Bill to Parliament.
Leahy who advised both bills noted that U.K. legislation pointed to self-improvement recursive as an early precursor of superintelligence which “must be controlled and prevented.”
Leahy stated that “superintelligence is not an instrument.” It’s not even a weapon. “It’s an enemy.”

The flight data recorder from the Amazon Air Boeing 767-390 crashed plane paints an interesting picture and also raises some questions about the events that took place in the cockpit when the plane careened down the runway of Miami International Airport.
21 Air’s 230,000-pound cargo plane, which was operated by Amazon for the contractor, hurled off the runway, crashing into an SUV, a van of cleaning crew, an airport fence and concrete barriers, before stopping. The van was carrying five people who were all killed.
Jennifer Homendy, chair of the NTSB, announced Tuesday night that investigators will be interviewing the 21 Air assistant chief pilot and the two pilots who crashed the plane on Wednesday. NTSB will also share information about what the pilots were saying in the minutes leading up to crash according to cockpit audio recordings.
Homendy, a spokesperson for the FBI in Washington said that investigators had recovered both the flight data and cockpit voice recorders – commonly referred to as “the black box” – found at the tail end of the jet. They were able to successfully access the data in their Washington lab.
The black box captured more than 400 data streams. Data released by the NTSB on Tuesday evening shows an odd sequence of events that led up to the aircraft going off the runway.
Chihoon Shin said that 30 seconds before the end of the recording, when the plane reached speeds exceeding 180 mph two landing wheels, the nose and back wheels, touched down onto the runway.
Nine seconds after the landing, the third set of gear on the left-rear of the aircraft touched down. The back wheels of the aircraft are always the first to touch down.
The plane had reached nearly 140mph when the 15 seconds recording began. To compensate, the brakes of the wheels on each wheel were removed and the throttles were set to maximum take-off. The brakes and throttles are then re-applied four seconds after the recording ends.
This sequence of events is not entirely clear. It’s not clear why this sequence of events occurred.
Shin, who spoke on Tuesday, said that “we need to talk to the crew to learn how they are manipulating the controls to better understand the situation.” Interviews with the crew will fill in any blanks we may have in regards to actions they took but did not express. We need to understand this completely before we can analyze the events that occurred with throttles.
The NTSB stated that there was no evidence of speed brakes being used or thrust reversers deployed at the time. The two systems can help slow down a plane. Speed brakes use aerodynamic surfaces, and thrust reversers divert air away from the engine.
The cockpit voice recorder captured more than two hours worth of audio, but NTSB wants to speak with the pilots first before disclosing what they said. Miami-Dade Sheriff’s Office has identified the two pilots involved as Joseph Carol, and Jaime Felipe Silva Molina.
The NTSB reported that Carol, a 55-year old captain, was certified as a Boeing 767 pilot since May. Silva Molina had over 2,500 flying hours and was certified to fly the Boeing 767 in April 2025.
Investigators from the NTSB collected evidence on Tuesday. They took measurements, inspected the cockpit and engines, and found debris in the parking lot where the aircraft stopped. The black box of one vehicle that was struck was recovered.
The NTSB visited 21 Air’s Miami operations center and spoke with the dispatcher of the flight who was responsible for pre-flight preparation and tracking the flight from San Juan, to Miami.
NTSB reports that the plane carried more than 32 pounds of cargo, mainly contact lenses.
The investigators visited both the terminal radar approach control and the air traffic tower, where planes are guided into the airport.
The investigators will enter the other areas of the plane on Wednesday to collect additional evidence. The NTSB reported that one runway at Miami International Airport is still closed. However, a second runway was opened for departures Tuesday night.
The investigators’ focus is on gathering perishable evidence at the accident scene. Homendy explained that the length of time gives them a chance to get it right.
Information that is received within hours or days after an accident may be incorrect. It takes time for us to verify that the information we provide is accurate,” she said Tuesday. We have a lot to do and we need to verify a great deal of information. It takes time to validate all the information.
Company says that the 5 victims are like family
Miami-Dade sheriff Rosie Cordero Stutz confirmed Tuesday that the five victims of this van accident all worked at Professional Ocean Service Corp.
According to the sheriff’s office, victims included Rolando Aleman Leon (55), Yoel Rodanjo (53) Julio C. Pineda (75) Carlos Acosta Fajardo (53), and Javierkys Reyes Quevedo (47)
Professional Ocean Service Corp. expressed its “heartbreak” at this tragedy, and said that their employees are like family.
The company expressed its deepest sympathies to all employees and their families whose lives were forever altered. We grieve for those who have lost loved ones and we pray that those injured will recover and heal.
The sheriff reported that three people who were injured in the accident remained hospitalized on Tuesday morning.
The sheriff’s department reported that Ridoel Diaz, 32 years old, was an occupant in the van and Eugenio Coronador, aged 79 in the vehicle next to it, both were critically injured. The sheriff’s department reported that Roosevelt Sebastian Perdomo Torres (36), who was in van, is in stable condition.
Homendy stated that the van and SUV were both struck by the plane on an airport service route, while the plane also hit the SUV outside of an airport fence.
This report was contributed by Andy Rose of CNN, Maria Aguilar Prieto from CNN, Julianna Brgg, Chris Boyette and Amanda Musa. Alexandra Skores also contributed.

What happened? Meta’s (META), stock rose more than 5% Wednesday.
Why the change? Shares in the company rose after it debuted Muse – an AI assistant. Wall Street analysts have high hopes for the chatbot.
In a note to clients, Mizuho Securities’ analysts stated that they believe Meta’s Muse AI consumer agent is the start of an important product cycle. This has not yet been priced into Meta’s shares. It has an outperform rating on the stock and set a price of $750 for it.
KeyBanc Capital Markets analyst have assigned the stock an Overweight ranking with a $780 price target. They wrote, “continued to believe that the market undervalues Meta’s AI product cycle and positioning.”
Meta, on Tuesday, said that you should also know the following: Muse is “designed around how people communicate already, so messaging it in the Muse application or in WhatsApp works the same as messaging someone else.”
Mark Zuckerberg, the CEO of Facebook, praised in a video his tool’s privacy- and security-related capabilities.
Muse cannot read your credit card numbers or passwords because we have built a Secure Credential Store. Muse checks with you directly before sensitive actions such as making payments or sending out messages”, Zuckerberg stated.
He added, “We are modeling our encryption after what we have built for WhatsApp where Meta cannot see your messages.”
Meta shares are back to where they were earlier in the year. The stock has lost 1% so far this year.
Ines Ferre, a Senior Business Reporter for Yahoo Finance.
Here you can find a detailed analysis of recent stock market events and news.