Good morning. Andrew here. Breaking: Doug McMillon, Walmart’s C.E.O., will retire and be succeeded by John Furner, who has been with the company for more than 30 years. We often talk about the transformative possibilities of artificial intelligence. But we’ve got some news this morning that is disturbing and important: Anthropic revealed that Chinese-sponsored hackers used its software to attack a group of companies and government agencies. More details on the breach — and its implications — below. We also dive into why the stock market is selling off and the latest bet by Michael Burry, the contrarian investor of “The Big Short” fame. Hint: It’s about A.I. Speaking of A.I., we’ll discuss its future with Dario Amodei of Anthropic and Alex Karp of Palantir at the DealBook Summit on Dec. 3. We hope you’ll join us. (Was this newsletter forwarded to you? Sign up here.)
A new era of hackingUnderlying the hundreds of billions being spent on artificial intelligence worldwide is the proposition that the technology will eventually bring transformative prosperity. But a stunning new disclosure by Anthropic underscores the flip side of the A.I. race: The technology can also be weaponized to a potentially devastating degree. Chinese state-sponsored hackers used Anthropic’s Claude tools in unprecedented ways, the A.I. start-up said yesterday. The twist is that the perpetrators used Claude’s “agentic” features — which let A.I. take action on behalf of users — to execute the attacks.
Anthropic said it eventually detected the campaign, shut down accounts used in the attack, notified the targets and strengthened Claude’s guardrails. A.I. did 80 percent to 90 percent of the work, Anthropic said. A.I. software is already being used in hacks, including by China, Russia and other governments. But this campaign was done “literally with the click of a button,” Jacob Klein, the company’s head of threat intelligence, told The Wall Street Journal. It was able to make “thousands of requests per second,” a rate that’s “simply impossible” for humans to match. The scheme wasn’t perfect: Claude hallucinated some security credentials and claimed some documents were pilfered instead of publicly available. But it underscores the power of A.I. agents, a feature that’s a big priority of Anthropic, OpenAI, Microsoft, Google and others. Anthropic’s response to the incident is to keep advancing its software. “If we don’t enable defenders to have a very substantial permanent advantage, I’m concerned that we maybe lose this race,” Logan Graham, another Anthropic security executive, told The Journal. Martin Peers of The Information had a snarky take: “That’s a little like the gun lobby claiming we shouldn’t ban guns because people need weapons for personal defense!” The incident raises other concerns:
Artificial intelligence start-ups dominate funding talks. Elon Musk’s xAI is raising $15 billion in fresh capital, according to CNBC. (Elon Musk has denied the report). Thinking Machines, which was founded by the former OpenAI executive Mira Murati, is in early talks to raise money at a $50 billion valuation, Bloomberg reports. And the company behind Cursor, the A.I.-based coding tool that has become indispensable in Silicon Valley, has raised $2.3 billion in its latest round, at a $29.3 billion valuation. Blue Origin lands a used rocket back on Earth. A New Glenn rocket booster from the company founded by Jeff Bezos landed on a floating platform upon its terrestrial return, something Elon Musk’s SpaceX first did a decade ago. Doing that a second consecutive time allows Blue Origin to compete for more commercial and government space launches. The White House prepares tariff exemptions in an effort to lower food prices. The Trump administration is preparing a proposal to exempt imports of foods such as beef and citrus products from some of President Trump’s steepest tariffs, announced in April. The list of products is not final, The Times reports, but the potential exemptions come as inflation becomes a growing political issue and weighs on the president’s approval rating. Why markets are tumblingBitcoin, global stocks, the dollar: All are selling off again today after the S&P 500 suffered its biggest single-day loss in a month yesterday. Here is what’s behind the market swoon. Economic indicator concerns are worrying investors. The Bureau of Labor Statistics will probably release the September jobs report next week, analysts say. The data, which was supposed to have been published on Oct. 3, should give Wall Street and the Fed a better grasp on hiring. But things get messy after that. Kevin Hassett, one of President Trump’s top economic advisers, said yesterday that the October nonfarm payrolls report will be released eventually too, but without the unemployment rate. That might cloud Fed officials’ understanding of how the labor market fared during much of the government shutdown when thousands of workers were furloughed or didn’t collect pay. And the chances of getting timely reads on inflation — especially the Consumer Price Index reports for October and November — any time soon are unclear, as data collection was most likely disrupted by the shutdown. That leaves the Fed in a bind. Alberto Musalem, the St. Louis Fed president, is the latest voting member of the central bank to express fears that lowering borrowing costs might stoke inflation. “We need to proceed and tread with caution,” he said yesterday. The probability that the Fed will cut rates at its meeting next month has narrowed in recent days. According to CME Group, futures traders penciled in roughly 50 percent odds this morning, down from nearly 60 percent on Wednesday. That’s unnerving investors. There are also stock valuation jitters. High-flying tech stocks, including those at the forefront of the artificial intelligence boom, were among the hardest-hit in yesterday’s market plunge. Shares in Tesla, Oracle and Palantir tumbled sharply as more investors question tech companies’ huge spending commitments and stratospheric valuations. These issues are likely to dominate market chatter for days, heading into the most-watched event of next week: the latest earnings report by Nvidia, the biggest driver of the A.I. boom. “The Big Short,” the sequel?Stepping into the debate over high stock valuations is none other than Michael Burry, the contrarian investor of “The Big Short” fame. His prescient bet against the U.S. housing market before its 2008 collapse was the stuff of Wall Street lore. Now he is taking on a new frothy corner of the market — some of the darlings of the artificial intelligence boom, writes Brian O’Keefe. A recap: Burry got Wall Street buzzing when his hedge fund, Scion Asset Management, revealed he’d made bets against two A.I. powerhouses. Scion’s latest 13F disclosure showed that Burry bought $9.2 million worth of put options against Palantir, with a notional value of $912 million. A similar bet against the A.I. chip giant Nvidia had a value of $187 million. “Sometimes, we see bubbles,” Burry wrote in an Oct. 30 post on X, his first post since 2023. Burry’s bet against Palantir underscores a wider Wall Street debate over A.I. stock valuations. Alex Karp, Palantir’s C.E.O., called Burry’s move “crazy” and suggested that he was engaging in market manipulation in a CNBC interview last week. Shorting Palantir has not been a winning strategy. Shares in the company are up more than 125 percent this year. And short sellers are down 82 percent betting against it through Nov. 11, according to S3 Partners, a market data firm. Burry may not be done there. In a post on X this week, he took aim at Meta and Oracle, and how they are accounting for depreciation in the context of their huge investments in Nvidia chips. “More detail coming November 25th. Stay tuned,” he added. (Burry did not respond to a request for an interview.) The stakes are high for Burry. He deregistered Scion this week, and will return investors’ money. Closing the fund suggests he will go it alone. “If you’re going to make these kinds of bets that are out of step with mainstream thinking, it’s probably easier just to do them in private,” Steve Sosnick, the chief strategist at Interactive Brokers, told DealBook. “Everything gets delivered to their house from burritos to medicine. Why not cash?”— Deepak Rao, vice president and manager of Robinhood Money, on a new service that will allow Robinhood bank account holders to get cash withdrawals delivered to their door. The service, which is rolling out first in New York, carries a delivery charge of up to $6.99.
Talking A.I. with the head of Williams F1 RacingEvery week, we’re asking a leader how he or she uses artificial intelligence. James Vowles, who leads the Formula One team Atlassian Williams Racing, told DealBook that the technology helps crack coded language that may signal what competitors will do next. His answers have been condensed and edited. How do you personally use A. I.? In Oxford, England, where I live, they pick up different colors of trash bins on different days. I wrote an agent that effectively scrapes the website and sends that to me the night before. If I can find something that automatically carries the bin out, then my life is complete. But I’m not there yet. Have you given any directives to your team about what you want them to do with A. I.? Each car is producing about 50,000 channels of data. And there are two cars. We’re consuming a large amount of video information from our competitors, probably 20 or 30 video feeds. There are audio feeds as well from every driver that’s out there. We can hear what they’re talking about. At the rate we’re growing data, it’s impossible to keep up with human beings. How do you pull insights from that data? An example: While teams talk in code on the radio about making a pit stop, you can basically infer it from the information you’re gathering. With A.I., what you have is the benefit of history: Let’s scrape through the last 20 races. We have a data stream for where they actually stopped in those races. Now let’s listen to what language was being used in the last minute before it happened. It doesn’t take long to understand it. We hope you’ve enjoyed this newsletter, which is made possible through subscriber support. Subscribe to The New York Times.
Deals
Politics, policy and regulation
Best of the rest
Correction: Yesterday’s DealBook newsletter incorrectly characterized ISS’s stance on offering “house” view on corporate votes. The company will continue to provide a benchmark voting policy; it will not stop offering them. Thanks for reading! We’ll see you tomorrow. We’d like your feedback. Please email thoughts and suggestions to dealbook@nytimes.com.
|
