This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Welcome to the dark side of crypto’s permissionless dream Jean-Paul Thorbjornsen, an Australian man in his mid-30s, with a rural Catholic upbringing, is a founder of THORChain, a blockchain through which users can swap one cryptocurrency for another and earn fees from making those swaps.THORChain is permissionless, so anyone can use it without getting prior approval from a centralized authority. As a decentralized network, the blockchain is built and run by operators located across the globe. During its early days, Thorbjornsen himself hid behind the pseudonym “leena” and used an AI-generated female image as his avatar. But around March 2024, he revealed his true identity as the mind behind the blockchain. More or less. If there is a central question around THORChain, it is this: Exactly who is responsible for its operations? It matters because in January last year, its users lost more than $200 million worth of their cryptocurrency in US dollars after THORChain transactions and accounts were frozen by a singular admin override, which users believed was not supposed to be possible given the decentralized structure.Thorbjornsen insists THORChain is helping realize bitcoin’s original purpose of enabling anyone to transact freely outside the reach of purportedly corrupt governments. Yet the network’s problems suggest that an alternative financial system might not be much better. Read the full story.
—Jessica Klein
The robots who predict the future To be human is, fundamentally, to be a forecaster. Occasionally a pretty good one. Trying to see the future, whether through the lens of past experience or the logic of cause and effect, has helped us hunt, avoid being hunted, plant crops, forge social bonds, and in general survive in a world that does not prioritize our survival. Today, we are awash in a sea of predictions so vast and unrelenting that most of us barely even register them. People’s desire for reliable forecasting is understandable. Still, nobody signed up for an omnipresent, algorithmic oracle mediating every aspect of their life. A trio of new books tries to make sense of our future-­focused world—how we got here, and what this change means. Each has its own prescriptions for navigating this new reality, but they all agree on one thing: Predictions are ultimately about power and control. Read the full story. —Bryan Gardiner These stories are both from the next print issue of MIT Technology Review magazine, which is all about crime. If you haven’t already, subscribe now to receive future issues once they land.  MIT Technology Review Narrated: Stratospheric internet could finally start taking off this year Today, an estimated 2.2 billion people still have either limited or no access to the internet, largely because they live in remote places. But that number could drop this year, thanks to tests of stratospheric airships, uncrewed aircraft, and other high-altitude platforms for internet delivery.This is our latest story to be turned into a MIT Technology Review Narrated podcast, which we’re publishing each week on Spotify and Apple Podcasts. Just navigate to MIT Technology Review Narrated on either platform, and follow us to get all our new content as it’s released.

The must-reads I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology. 1 Mark Zuckerberg is due to give evidence in a major social media addiction trialHe’ll face questioning over whether Meta does enough to protect young users. (CNN)2 Perplexity has abandoned ads inside its chatbot responsesBecause advertising can erode trust in AI, it reasons. (FT $)+ It’s a pretty big U-turn considering its previous stance. (The Verge)3 The US is being battered by a range of wild weatherFrom critical wildfire risks in some states, to winter storms in others. (WP $)4 Microsoft plans to spend $50 billion bringing AI to the Global South by 2030India is one of the fastest growing markets for the technology. (Reuters)+ One native startup has announced a new AI model for 22 Indian languages. (Bloomberg $)+ Inside India’s scramble for AI independence. (MIT Technology Review) 5 AI-powered private schools are failing studentsModels are being used to generate faulty lesson plans. (404 Media) 6 Land owners are selling out to data center buildersLand previously earmarked for housing is being sold off to the highest bidder. (WSJ $) 7 Tesla has agreed to stop using the term “autopilot” in CaliforniaThe DMV had previously also questioned its use of “Full Self-Driving.” (SF Chronicle $) 8 A new weight-loss drug may work a little too wellParticipants in a trial are dropping out at a much higher rate than normal. (NYT $)+ Intermittent fasting may not help us to shed the pounds after all. (New Scientist $)+ What we still don’t know about weight-loss drugs. (MIT Technology Review)
9 Is anyone still using Grindr?Bots and AI have rendered it virtually unusable for some people. (Vox)10 How to hack your dreamsNeuroscientists are figuring out new ways to influence what we dream about. (New Scientist $)+ I taught myself to lucid dream. You can too. (MIT Technology Review)
Quote of the day “I voted for this administration and didn’t really think about [AI] until it started to affect me.” —Lisa Garrett, a grandmother living in the city of Independence, Missouri, reflects on the Trump administration’s decision to embrace AI, the Financial Times reports. One more thing Hydrogen trains could revolutionize how Americans get aroundLike a mirage speeding across the dusty desert outside Pueblo, Colorado, the first hydrogen-fuel-cell passenger train in the United States is getting warmed up on its test track. It will soon be shipped to Southern California, where it is slated to carry riders on San Bernardino County’s Arrow commuter rail service before the end of the year.The best way to decarbonize railroads is the subject of growing debate among regulators, industry, and activists. The debate is partly technological, revolving around whether hydrogen fuel cells, batteries, or overhead electric wires offer the best performance for different railroad situations. But it’s also political: a question of the extent to which decarbonization can, or should, usher in a broader transformation of rail transportation.In the insular world of railroading, this hydrogen-powered train is a Rorschach test. To some, it represents the future of rail transportation. To others, it looks like a big, shiny distraction. Read the full story.
—Benjamin Schneider We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or skeet ’em at me.) + How to quickly declutter your home by being brutally honest with yourself.+ The filming locations for A Knight of the Seven Kingdoms are pretty breathtaking.+ Why a unicyclist decided to start juggling flaming torches in the middle of a Colorado pedestrian crossing is anyone guess, but good luck to him.+ How pepper took over the world (deservedly)

To be human is, fundamentally, to be a forecaster. Occasionally a pretty good one. Trying to see the future, whether through the lens of past experience or the logic of cause and effect, has helped us hunt, avoid being hunted, plant crops, forge social bonds, and in general survive in a world that does not prioritize our survival. Indeed, as the tools of divination have changed over the centuries, from tea leaves to data sets, our conviction that the future can be known (and therefore controlled) has only grown stronger.  Today, we are awash in a sea of predictions so vast and unrelenting that most of us barely even register them. As I write this sentence, algorithms on some remote server are busy trying to guess my next word based on those I have already typed. If you’re reading this online, a separate set of algorithms has likely already served you an ad deemed to be one you are most likely to click. (To the die-hards reading this story on paper, congratulations! You have escaped the algorithms … for now.) If the thought of a ubiquitous, mostly invisible predictive layer secretly grafted onto your life by a bunch of profit-hungry corporations makes you uneasy … well, same here. So how did all this happen? People’s desire for reliable forecasting is understandable. Still, nobody signed up for an omnipresent, algorithmic oracle mediating every aspect of their life. A trio of new books tries to make sense of our future-­focused world—how we got here, and what this change means. Each has its own prescriptions for navigating this new reality, but they all agree on one thing: Predictions are ultimately about power and control. The Means of Prediction: How AI Really Works (and Who Benefits)Maximilian KasyUNIVERSITY OF CHICAGO PRESS, 2025 In The Means of Prediction: How AI Really Works (and Who Benefits), the Oxford economist Maximilian Kasy explains how most predictions in our lives are based on the statistical analysis of patterns in large, labeled data sets—what’s known in AI circles as supervised learning. Once “trained” on such data sets, algorithms for supervised learning can be presented with all kinds of new information and then deliver their best guess as to some specific future outcome. Will you violate your parole, pay off your mortgage, get promoted if hired, perform well on your college exams, be in your home when it gets bombed? More and more, our lives are shaped (and, yes, occasionally shortened) by a machine’s answer to these questions.
If the thought of a ubiquitous, mostly invisible predictive layer secretly grafted onto your life by a bunch of profit-hungry corporations makes you uneasy … well, same here. This arrangement is leading to a crueler, blander, more instrumentalized world, one where life’s possibilities are foreclosed, age-old prejudices are entrenched, and everyone’s brain seems to be actively turning into goo. It’s an outcome, according to Kasy, that was entirely predictable.  AI adherents might frame those consequences as “unintended,” or mere problems of optimization and alignment. Kasy, on the other hand, argues that they represent the system working as intended. “If an algorithm selecting what you see on social media promotes outrage, thereby maximizing engagement and ad clicks,” he writes, “that’s because promoting outrage is good for profits from ad sales.” The same holds true for an algorithm that nixes job candidates “who are likely to have family-care responsibilities outside the workplace,” and the ones that “screen out people who are likely to develop chronic health problems or disabilities.” What’s good for a company’s bottom line may not be good for your job-hunting prospects or life expectancy.
Where Kasy differs from other critics is that he doesn’t think working to create less biased, more equitable algorithms will fix any of this. Trying to rebalance the scales can’t change the fact that predictive algorithms rely on past data that’s often racist, sexist, and flawed in countless other ways. And, he says, the incentives for profit will always trump attempts to eliminate harm. The only way to counter this is with broad democratic control over what Kasy calls “the means of prediction”: data, computational infrastructure, technical expertise, and energy.   A little more than half of The Means of Prediction is devoted to explaining how this might be accomplished—through mechanisms including “data trusts” (collective public bodies that make decisions about how to process and use data on behalf of their contributors) and corporate taxing schemes that try to account for the social harm AI inflicts. There’s a lot of economist talk along the way, about how “agents of change” might help achieve “value alignment” in order to “maximize social welfare.” Reasonable, I guess, though a skeptic might point out that Kasy’s rigorous, systematic approach to building new public-serving institutions comes at a time when public trust in institutions has never been lower. Also, there’s the brain goo problem.  To his credit, Kasy is a realist here. He doesn’t presume that any of these proposals will be easy to implement. Or that it will happen overnight, or even in the near future. The troubling question at the end his book is: Do we have that kind of time? Reading Kasy’s blueprint for seizing control of the means of prediction raises another pressing question. How on earth did we reach a point where machine-mediated prediction is more or less inescapable? Capitalism, might be Marx’s pithy response. Fine, as far as it goes, but that doesn’t explain why the same kinds of algorithms that currently model climate change are for some reason also deciding whether you get a new kidney or I get a car loan. The Irrational Decision: How We Gave Computers the Power to Choose for UsBenjamin RechtPRINCETON UNIVERSITY PRESS, 2026 If you ask Benjamin Recht, author of The Irrational Decision: How We Gave Computers the Power to Choose for Us, he’d likely tell you our current predicament has a lot to do with the idea and ideology of decision theory—or what economists call rational choice theory. Recht, a polymathic professor in UC Berkeley’s Department of Electrical Engineering and Computer Science, prefers the term “mathematical rationality” to describe the narrow, statistical conception that stoked the desire to build computers, informed how they would eventually work, and influenced the kinds of problems they would be good at solving.  This belief system goes all the way back to the Enlightenment, but in Recht’s telling, it truly took hold at the tail end of World War II. Nothing focuses the mind on risk and quick decision-making like war, and the mathematical models that proved especially useful in the fight against the Axis powers convinced a select group of scientists and statisticians that they might also be a logical basis for designing the first computers. Thus was born the idea of a computer as an ideal rational agent, a machine capable of making optimal decisions by quantifying uncertainty and maximizing utility. Intuition, experience, and judgment gave way, says Recht, to optimization, game theory, and statistical prediction. “The core algorithms developed in this period drive the automated decisions of our modern world, whether it be in managing supply chains, scheduling flight times, or placing advertisements on your social media feeds,” he writes. In this optimization-­driven reality, “every life decision is posed as if it were a round at an imaginary casino, and every argument can be reduced to costs and benefits, means and ends.” Today, mathematical rationality (wearing its human skin) is best represented by the likes of the pollster Nate Silver, the Harvard psychologist Steven Pinker, and an assortment of Silicon Valley oligarchs, says Recht. These are people who fundamentally believe the world would be a better place if more of us adopted their analytic mindset and learned to weigh costs and benefits, estimate risks, and plan optimally. In other words, these are people who believe we should all make decisions like computers. 

How might we demonstrate that (unquantifiable) human intuition, morality, and judgment are better ways of addressing some of the world’s most important and vexing problems? It’s a ridiculous idea for multiple reasons, he says. To name just one, it’s not as if humans couldn’t make evidence-based decisions before automation. “Advances in clean water, antibiotics, and public health brought life expectancy from under 40 in the 1850s to 70 by 1950,” Recht writes. “From the late 1800s to the early 1900s, we had world-changing scientific breakthroughs in physics, including new theories of thermodynamics, quantum mechanics, and relativity.” We also managed to build cars and airplanes without a formal system of rationality and somehow came up with societal innovations like modern democracy without optimal decision theory.  So how might we convince the Pinkers and Silvers of the world that most decisions we face in life are not in fact grist for the unrelenting mill of mathematical rationality? Moreover, how might we demonstrate that (unquantifiable) human intuition, morality, and judgment might be better ways of addressing some of the world’s most important and vexing problems? Prophecy: Prediction, Power, and the Fight for the Future, from Ancient Oracles to AICarissa VélizDOUBLEDAY, 2026 One might start by reminding the rationalists that any prediction, computational or otherwise, is really just a wish—but one with a powerful tendency to self-fulfill. This idea animates Carissa Véliz’s wonderfully wide-ranging polemic Prophecy: Prediction, Power, and the Fight for the Future, from Ancient Oracles to AI.  A philosopher at the University of Oxford, Véliz sees a prediction as “a magnet that bends reality toward itself.” She writes, “When the force of the magnet is strong enough, the prediction becomes the cause of its becoming true.”  Take Gordon Moore. While he doesn’t come up in Prophecy, he does figure somewhat prominently in Recht’s history of mathematical rationality. A cofounder of the tech giant Intel, Moore is famous for his 1965 prediction that the density of transistors in integrated circuits would double every two years. “Moore’s Law” turned out to be true, and remains true today, although it does seem to be running out of steam thanks to the physical size limits of the silicon atom. One story you can tell yourself about Moore’s Law is that Gordon was just a prescient guy. His now-classic 1965 opinion piece “Cramming More Components onto Integrated Circuits,” for Electronics magazine, simply extrapolated what computing trends might mean for the future of the semiconductor industry.  Another story—the one I’m guessing Véliz might tell—is that Moore put an informed prediction out into the world, and an entire industry had a collective interest in making it come true. As Recht makes clear, there were and remain obvious financial incentives for companies to make faster and smaller computer chips. And while the industry has likely spent billions of dollars trying to keep Moore’s Law alive, it’s undoubtedly profited even more from it. Moore’s Law was a helluva strong magnet.  Predictions don’t just have a habit of making themselves come true, says Véliz. They can also distract us from the challenges of the here and now. When an AI boomer promises that artificial general intelligence will be the last problem humanity needs to solve, it not only shapes how we think about AI’s role in our lives; it also shifts our attention away from the very real and very pressing problems of the present day—problems that in many cases AI is causing.
In this sense, the questions around predictions (Who’s making them? Who has the right to make them?) are also fundamentally about power. It’s no accident, Véliz says, that the societies that rely most heavily on prediction are also the ones that tend toward oppression and authoritarianism. Predictions are “veiled prescriptive assertions—they tell us how to act,” she writes. “They are what philosophers call speech acts. When we believe a prediction and act in accordance with it, it’s akin to obeying an order.” As much as tech companies would like us to believe otherwise, technology is not destiny. Humans make it and choose how to use it … or not use it. Maybe the most appropriate (and human) thing we can do in the face of all the uninvited daily predictions in our lives is to simply defy them.  Bryan Gardiner is a writer based in Oakland, California.

“We’re out of airspace now. We can do whatever we want,” Jean-Paul Thorbjornsen tells me from the pilot’s seat of his Aston Martin helicopter. As we fly over suburbs outside Melbourne, Australia, it’s becoming clear that doing whatever he wants is Thorbjornsen’s MO.  Upper-middle-class homes give way to vineyards, and Thorbjornsen points out our landing spot outside a winery. People visiting for lunch walk outside. “They’re going to ask for a shot now,” he says, used to the attention drawn by his luxury helicopter, emblazoned with the tail letters “BTC” for bitcoin (the price tag of $5 million in Australian dollars—$3.5 million in US dollars today—was perhaps reasonable for someone who claims a previous crypto project made more than AU$400 million, although he also says those funds were tied up in the company).  Thorbjornsen is a founder of THORChain, a blockchain through which users can swap one cryptocurrency for another and earn fees from making those swaps. THORChain is permissionless, so anyone can use it without getting prior approval from a centralized authority. As a decentralized network, the blockchain is built and run by operators located across the globe, most of whom use pseudonyms.  During its early days, Thorbjornsen himself hid behind the pseudonym “leena” and used an AI-generated female image as his avatar. But around March 2024, he revealed that he, an Australian man in his mid-30s, with a rural Catholic upbringing, was the mind behind the blockchain. More or less. 
If there is a central question around THORChain, it is this: Exactly who is responsible for its operations? Blockchains as decentralized as THORChain are supposed to offer systems that operate outside the centralized leadership of corruptible governments and financial institutions. If a few people have outsize sway over this decentralized network—one of a handful that operate at such a large scale—it’s one more blemish on the legacy of bitcoin’s promise, which has already been tarnished by capitalistic political frenzy.    Who’s responsible for THORChain matters because in January last year, its users lost more than $200 million worth of their cryptocurrency in US dollars after THORChain transactions and accounts were frozen by a singular admin override, which users believed was not supposed to be possible given the decentralized structure. When the freeze was lifted, some users raced to pull their money out. The following month, a team of North Korean hackers known as the Lazarus Group used THORChain to move roughly $1.2 billion of stolen ethereum taken in the infamous hack of the Dubai-based crypto exchange Bybit. 
Thorbjornsen explains away THORChain’s inability to stop the movement of stolen funds, or prevent a bank run, as a function of its decentralized and permissionless nature. The lack of executive powers means that anyone can use the network for any reason, and arguably there’s no one to hold accountable when even the worst goes down. But when the worst did go down, nearly everyone in the THORChain community, and those paying attention to it in channels like X, pointed their fingers at Thorbjornsen. A lawsuit filed by the THORChain creditors who lost millions in January 2025 names him. A former FBI analyst and North Korea specialist, reflecting on the potential repercussions for helping move stolen funds, told me he wouldn’t want to be in Thorbjornsen’s shoes. THORChain was designed to make decisions based on votes by node operators, where two-thirds majority rules. That’s why I traveled to Australia—to see if I could get a handle on where he sees himself and his role in relation to the network he says he founded. According to Thorbjornsen, he should not be held responsible for either event. THORChain was designed to make decisions based on votes by node operators—people with the computer power, and crypto stake, to run a cluster of servers that process the network’s transactions. In those votes, a two-thirds majority rules. Then there’s the permissionless part. Anyone can use THORChain to make swaps, which is why it’s been a popular way for widely sanctioned entities such as the government of North Korea to move stolen money. This principle goes back to the cypherpunk roots of bitcoin, a currency that operates outside of nation-states’ rules. THORChain is designed to avoid geopolitical entanglements; that’s what its users like about it. But there are distinct financial motivations for moving crypto, stolen or not: Node operators earn fees from the funds running through the network. In theory, this incentivizes them to act in the network’s best interests—and, arguably, Thorbjornsen’s interests too, as many assume his wealth is tied to the network’s profits. (Thorbjornsen says it is not, and that it comes instead from “many sources,” including “buying bitcoin back in 2013.”) Now recent events have raised critical questions, not just about Thorbjornsen’s outsize role in THORChain’s operations, but also about the blockchain’s underlying nature. If THORChain is decentralized, how was a single operator able to freeze its funds a month before the Bybit hack? Could someone have unilaterally decided to stop the stolen Bybit funds from coming through the network, and chosen not to? 

Thorbjornsen insists THORChain is helping realize bitcoin’s original purpose of enabling anyone to transact freely outside the reach of purportedly corrupt governments. Yet the network’s problems suggest that an alternative financial system might not be much better. Decentralized?  On February 21, 2025, Bybit CEO Ben Zhou got an alarming call from the company’s chief financial officer. About $1.5 billion US of the exchange’s ethereum token, ETH, had been stolen.  The FBI attributed the theft to the Lazarus Group. Typically, criminals will want to convert ETH to bitcoin, which is much easier to convert in turn to cash. Knowing this, the FBI issued a public service announcement on February 26 to “exchanges, bridges … and other virtual asset service providers,” encouraging them to block transactions from accounts related to the hack.  Someone posted that announcement in THORChain’s private, invite-only developer channel on Discord, a chat app used widely by software engineers and gamers. While other crypto exchanges and bridges (which facilitate transactions across different blockchains) heeded the warning, THORChain’s node operators, developers, and invested insiders debated about whether or not to close the trading gates, a decision requiring a majority vote. “Civil war is a very strong term, but there was a strong rift in the community,” says Boone Wheeler, a US-based crypto enthusiast. In 2021, Wheeler purchased some rune, THORChain’s Norse-mythology-themed native token, and he has been paid to write articles about the network to help advertise it. The rift formed “between people who wanted to stay permissionless,” he says, “and others who wanted to blacklist the funds.” Wheeler, who says he doesn’t run a node or code for THORChain, fell on the side of remaining permissionless. However, others spoke up for blocking the transfers. THORChain, they argued, wasn’t decentralized enough to keep those running the network safe from law enforcement—especially when those operators were identifiable by their IP addresses, some based in the US. “We are not the morality police,” someone with the username @Swing_Pop wrote on February 27 in the developer Discord. THORChain’s design includes up to 120 nodes whose operators manage transactions on the network through a voting process. Anyone with hosting hardware can become an operator by funding nodes with rune as collateral, which provides the network with liquidity. Nodes can respond to a transaction by validating it or doing nothing. While individual transactions can’t be blocked, trading can be halted by a two-thirds majority vote. 
A team of North Korean hackers used THORChain to move roughly $1.2 billion of ethereum stolen from the crypto exchange Bybit. Nodes are also penalized for not participating in voting, which the system labels as “bad behavior.” Every 2.5 days, THORChain automatically “churns” nodes out to ensure that no one node gains too much control. The nodes that chose not to validate transactions from the Bybit hack were automatically “churned” out of the system. Thorbjornsen says about 20 or 30 nodes were booted from the network in this way. (Node operators can run multiple nodes, and 120 are rarely running simultaneously; at the time of writing, 55 unique IDs operated 103 nodes.) By February 27, some node operators were prepared to leave the network altogether. “It’s personally getting beyond my risk tolerance,” wrote @Runetard in the dev Discord. “Sorry to those of the community that feel otherwise. There are a bunch of us holding all the risk and some are getting ready to walk away.”
According to one estimate, THORChain earned between $5 million and $10 million from the heist. Even so, the financial incentive for the network operators who remained was significant. As one member of the dev Discord put it earlier that day, $3 million had been “extracted as commission” from the theft by those operating THORChain. “This is wrong!” they wrote. Thorbjornsen weighed in on this back-and-forth, during which nodes paused and unpaused the network. He now says there was a right and wrong way for node operators to have behaved. “The correct way of doing things,” he says, was for node operators who opposed processing stolen funds to “go offline and … get [themselves] kicked out” rather than try to police who could use THORChain. He also says that while operators could discuss stopping transactions, “there was simply no design in the code that allowed [them] to do that.” However, a since-deleted post from his personal X account on March 3, 2025, stated: “I pushed for all my nodes to unhalt trading [keep trading]. Threatened to yank bond if they didn’t comply. Every single one.” (Thorbjornsen says his social media team ran this account in 2025.)  In an Australian 7 News Spotlight documentary last June, Thorbjornsen estimated that THORChain earned between $5 million and $10 million from the heist. When asked in that same documentary if he received any of those fees, he replied, “Not directly.” When we spoke, I asked him to elaborate. He said he’s “not a recipient” of any funds THORChain sets aside for developers or marketers, nor does he operate any nodes. He was merely speaking generally, he told me: “All crypto holders profit indirectly off economic activity on any chain.” KAGAN MCLEOD Most important to Thorbjornsen was that, despite “huge pressure to shut the protocol down and stop servicing these swaps,” THORChain chugged along. He also notes that the hackers’ tactics, moving fast and splitting funds across multiple addresses, made it difficult to identify “bad swaps.” Blockchain experts like Nick Carlsen, a former FBI analyst at the blockchain intelligence company TRM Labs, don’t buy this assessment. Other services similar to THORChain were identifying and rejecting these transactions. Had THORChain done the same, Carlsen adds, the stolen funds could have been contained on the Ethereum network, which “would have basically denied North Korea the ability to kick off this laundering process.” 
And while THORChain touts its decentralization, in “practical applications” like the Lazarus Group’s theft, “most de-fi [decentralized finance] protocols are centralized,” says Daren Firestone, an attorney who represents crypto industry whistleblowers, citing a 2023 US Treasury Department risk assessment on illicit finance.  With centralization comes culpability, and in these cases, Firestone adds, that comes down to “who profits from [the protocol], so who creates it? But most importantly, who controls it?” Is there someone who can “hit an emergency off switch? … Direct nodes?” Many answer these questions with Thorbjornsen’s name. “Everyone likes to pass the blame,” he says, even though he wasn’t alone in building THORChain. “​​In the end, it all comes back to me anyway.” THORChain origins According to Thorbjornsen, his story goes like this. The third of 10 homeschooled children in a “traditional” Catholic household in rural Australia, he spent his days learning math, reading, writing, and studying the Bible. As he got older, he was also responsible for instructing his younger siblings. Wednesday was his day to move the solar panels that powered their home. His parents “installed” a mango and citrus orchard, more to keep nine boys busy than to reap the produce, he says. “We lived close to a local airfield,” Thorbjornsen says, “and I was always mesmerized by these planes.” He joined the Australian air force and studied engineering, but he says the military left him feeling like “a square peg in a round hole.” He adds that doing things his own way got him frequently “pulled aside” by superiors.
“That’s when I started looking elsewhere,” he says, and in 2013, he found bitcoin. It appealed because it existed “outside the system.” During the 2017 crypto bull run, Thorbjornsen raised AU$12 million in an initial coin offering for CanYa, a decentralized marketplace he cofounded. CanYa ultimately “died” in 2018, and Thorbjornsen pivoted to a “decentralized liquidity” project that would become THORChain. He worked with a couple of different developer teams, and then, in 2019, he clicked with an American developer, Chad Barraford, at a hackathon in Germany. Barraford (who declined to be interviewed for this story) was an early public face of THORChain.  Around this time, Thorbjornsen says, “a couple of us helped manage the payroll and early investment funds.” In a 2020 interview, Kai Ansaari, identified as a THORChain “project lead,” wrote, “We’re all contributors … There’s no real ‘lead,’ ‘CEO,’ ‘founder,’ etc.” In interviews conducted since he came out from behind the “leena” account in 2024, Thorbjornsen has positioned himself as a key lead. He now says his plan had always been to hand over the account, along with command powers and control of THORChain social media accounts, once the blockchain had matured enough to realize its promise of decentralization. In 2021, he says, he started this process, first by ceasing to use his own rune to back node operators who didn’t have enough to supply their own funding (this can be a way to influence node votes without operating a node yourself). That year, the protocol suffered multiple hacks that resulted in millions of dollars in losses. Nine Realms, a US-incorporated coding company, was brought on to take over THORChain’s development. Thorbjornsen says he passed “leena” over to “other community members” and “left crypto” in 2021, selling “a bunch of bitcoin” and buying the helicopter.  Despite this crypto departure, he came back onto the scene with gusto in 2024 when he revealed himself as the operator of the “leena” account. “​​For many years, I stayed private because I didn’t want the attention,” he says now.  By early 2024 Thorbjornsen considered the network to be sufficiently decentralized and began advertising it publicly. He started regularly posting videos on his TikTok and YouTube channels (“Two sick videos every week,” in the words of one caption) that showed him piloting his helicopter wearing shirts that read “Thor.” By November 2024, Thorbjornsen, who describes himself as “a bit flamboyant,” was calling himself THORChain’s CEO (“chief energy officer”) and the “master of the memes” in a video from Binance Blockchain Week, an industry conference in Dubai. You need “strong memetic energy,” he says in the video, “to create the community, to create the cult.” Cults imply centralized leadership, and since outing himself as “leena,” Thorbjornsen has publicly appeared to helm the project, with one interviewer deeming him the “THORChain Satoshi” (an allusion to the pseudonymous creator of bitcoin).  One consequence of going public as a face of the protocol: He’s received death threats. “I stirred it up. Do I regret it? Who knows?” he said when we met in Australia. “It’s caused a lot of chaos.”  But, he added, “this is the bed that I’ve laid.” When we spoke again, months later, he backtracked, saying he “got sucked into” defending THORChain in 2024 and 2025 because he was involved from 2018 to 2021 and has “a perspective on how the protocol operates.” Centralized?  Ryan Treat, a retired US Army veteran, woke up one morning in January 2025 to some disturbing activity on X. “My heart sank,” he says. THORFi, the THORChain program he’d used to earn interest on the bitcoin he’d planned to save for his retirement, had frozen all accounts—but that didn’t make sense. THORFi featured a lending and saving program said to give users “complete control” and self-custody of their crypto, meaning they could withdraw it at any time.  Treat was no crypto amateur. He bought his first bitcoin at around “$5 apiece,” he says, and had always kept it off centralized exchanges that would maintain custody of his wallets. He liked THORChain because it claimed to be decentralized and permissionless. “I got into bitcoin because I wanted to have government-less money,” he says.  We were told it was decentralized. Then you wake up one morning and read this guy had an admin mimir. Many who’d used THORFi lending and saving programs felt similarly. Users I interviewed differentiated THORChain from centralized lending platforms like BlockFi and Celsius, both of which offered extraordinarily high yields before filing for bankruptcy in 2022. “I viewed THORChain as a decentralized system where it was safer,” says Halsey Richartz, a Florida-based THORFi creditor, with “vanilla, 1% passive yield.” Indeed, users I spoke with hadn’t felt the need to monitor their THORFi deposits. “Only your key can be used to withdraw your funds,” the product’s marketing materials insisted. “Savers can withdraw their position to native assets at any time.” So on January 9, when the “leena” account announced that an admin key had been used to pause withdrawals, it took THORFi users by surprise—and seemed to contradict the marketing messaging around decentralization. “We were told that it was decentralized, and you wake up one morning and read an article that says ‘This guy, JP, had an admin mimir,’” says Treat, referring to Thorbjornsen, “and I’m like, ‘What the fuck is an admin mimir?’” The admin mimir was one of “a bunch of hard-coded admin keys built into the base code of the system,” says Jonathan Reiter, CEO of the blockchain intelligence company ChainArgos. Those with access to the keys had the ability to make executive decisions on the blockchain—a function many THORChain users didn’t realize could supersede the more democratic decisions made by node votes. These keys had been in THORChain’s code for years and “let you control just about anything,” Reiter adds, including the decision to pause the network during the hacks in 2021 that resulted in a loss of more than $16 million in assets.  Thorbjornsen says that one key was given to Nine Realms, while another was “shared around the original team.” He told me at least three people had them, adding, “I can neither confirm nor deny having access to that mimir key, because there’s no on-chain registry of the keys.” Regardless of who had access, Thorbjornsen maintains that the admin mimir mechanism was “widely known within the community, and heavily used throughout THORChain’s history” and that any action taken using the keys “could be largely overruled by the nodes.” Indeed, nodes voted to open withdrawals back up shortly after the admin key was used to pause them. By then, those burned by THORFi argue, the damage had already been done. The executive pause to withdrawals, for some, signaled that something was amiss with THORFi. This led to a bank run after the pause was lifted, until the nodes voted to freeze withdrawals permanently (which Thorbjornsen had suggested in a since-deleted post on X), separating users from crypto worth around $200 million in US dollars on January 23. THORFi users were then offered a token called TCY (THORChain Yield), which they could claim with the idea that, when its price rose to $1, they would be made whole. (The price, as of writing, sits at $0.16.) Who used the key? Thorbjornsen maintains he didn’t do it, but he claims he knows who did and won’t say. He says he’d handed over the “leena” account and doesn’t “have access to any of the core components of the system,” nor has he for “at least three years.” He implies that whoever controlled “leena” at the time used the admin key to pause network withdrawals. A video released by Nine Realms on February 20, 2025, names Thorbjornsen as the activator of the key, stating, “JP ended up pausing lenders and savers, preventing withdrawals so that we can work out … [a] payback plan on them.” Thorbjornsen told me the video was “not factual.” Multiple blockchain analysts told me it would be extremely difficult to determine who used the admin mimir key. A month after it was used to pause the network, THORChain said the key had been “removed from the network.” At least you can’t find the words “admin mimir” in THORChain’s base code; I’ve looked.  Culpability After the debacle of the THORFi withdrawal freeze, Richartz says, he tried to file reports with the Miami-Dade Police Department, the Florida Department of Law Enforcement, the FBI, the Securities and Exchange Commission, the Commodity Futures Trading Commission, the Federal Trade Commission, and Interpol. When we spoke in November, he still hadn’t been able to file with the city of Miami. They told him to try small claims court. “I was like, no, you don’t understand … a post office box in Switzerland is the company address,” he says. “It underscored to me how little law enforcement even knows about these crimes.”  As for the Bybit hack, at least one government has retaliated against those who facilitate blockchain projects. Last April German authorities shut down eXch, an exchange suspected of using THORChain to process funds Lazarus stole from Bybit, says Julia Gottesman, cofounder and head of investigations at the cybersecurity group zeroShadow. Australia, she adds, where Thorbjornsen was based, has been “slow to try to engage with the crypto community, or any regulations.” KAGAN MCLEOD In response to requests for comment, Australia’s Department of Home Affairs wrote that at the end of March 2026, the country’s regulatory powers will expand to include “exchanges between the same type of cryptocurrency and transfers between different types.” They did not comment on specific investigations. Crypto and finance experts disagree about whether THORChain engaged in money laundering, defined by the UN as “the processing of criminal proceeds to disguise their illegal origin.” But some think it fits the definition. Shlomit Wagman, a Harvard fellow and former head of Israel’s anti-money-laundering agency and its delegation to the Financial Action Task Force (FATF), thinks the Bybit activity was money laundering because THORChain helped the hackers “transfer the funds in an unsupervised manner, completely outside of the scope of regulated or supervised activity.”  And by helping with conversions, Carlsen says, THORChain enabled bad actors to turn stolen crypto into usable currency. “People like [Thorbjornsen] have a personal degree of culpability in sustaining the North Korean government,” he says. Thorbjornsen counters that THORChain is “open-source infrastructure.” Meanwhile, just days after the hack, Bybit issued a 10% bounty on any funds recovered. As of mid-January this year, between $100 million and $500 million worth of those funds in US dollars remain unaccounted for, according to Gottesman of zeroShadow, which was hired by Bybit to recover funds following the hack. Thorbjornsen hacked For Thorbjornsen, it’s just another day at the casino. That’s the comparison he made during his regrettable 7 News Spotlight interview about the Bybit heist, and he repeated it when we met. “You go to a casino, you play a few games, you expect to lose,” he told me. “When you do actually go to zero, don’t cry.” Thorbjornsen, it should be noted, has lost at the casino himself. In September, he says, he got a Telegram message from a friend, inviting him to a Zoom meeting. He accepted and participated in a call with people who had “American voices.” Ultimately, Thorbjornsen describes himself as a guy who’s had a bad year, fending off “threat vectors” left and right. After the meeting, Thorbjornsen learned that his friend’s Telegram had been hacked. Whoever was responsible had used the Zoom link to remotely install software on Thorbjornsen’s computer, which “got access to everything”—his email, his crypto wallets, a bitcoin-based retirement fund. It cost him at least $1.2 million. The blockchain sleuth known as ZachXBT traced the funds and attributed the hack to North Korea.  ZachXBT called it “poetic.” Ultimately, Thorbjornsen describes himself as a guy who’s had a bad year. He says he had to liquidate his crypto assets because he’s dealing with the fallout of a recent divorce. He also feels he is fending off “threat vectors” left and right. More than once, he asked if I was a private investigator masquerading as a journalist. Still, his many contradictions don’t inspire confidence. He doesn’t have any more crypto assets, he says. However, the crypto wallet he shared with me so I could pay him back for lunch showed that it contained assets worth more than $143,000 in US dollars. He now says it wasn’t his wallet. He says he doesn’t control THORChain’s social media, but he’d also told me that he runs the @THORChain X account (later backtracking and saying the account is “delegated” to him for trickier questions). He insists that he does not care about money. He says that in the robot future, the AI-powered hive mind will become our benevolent overlord, rendering money obsolete, so why give it much thought? Yet as we flew back from the vineyard, he pointed out his new house from the helicopter. It resembles a compound. He says he lives there with his new wife.  Multiple people I spoke with about Thorbjornsen before I met him warned me he wasn’t trustworthy, and he’s undeniably made fishy statements. For instance, the presence of a North Korean flag in a row of decals on the tail of his helicopter suggested an affinity with the country for which THORChain has processed so much crypto. Thorbjornsen insists he had requested the flag of Australia’s Norfolk Island, calling the mix-up “a complete coincidence.” The flags were gone by the time of our flight, apparently removed during a recent repair. “Money is a meme,” he says. “Money does not exist.” Meme or not, it’s had real repercussions for those who have interacted with THORChain, and those who wound up losing have been looking for someone to blame.  When I spoke with Thorbjornsen again in January, he appeared increasingly concerned that he is that someone. He’s spending more time in Singapore, he told me. Singapore happens to have historically denied extraditions to the US for money-laundering prosecutions.  Jessica Klein is a Philadelphia-based freelance journalist covering intimate partner violence, cryptocurrency, and other topics.

Introducing our National Partnerships for AI and collaboration in IndiaWe believe AI will be the most transformative technology in human history and that it should be deployed in ways that benefit all of humanity. This requires deep, strategic collaboration between frontier AI labs, governments, academia, and civil society.To fully realise AI’s potential, Google DeepMind is working with governments through our National Partnerships for AI initiative to broaden access to our frontier AI capabilities, helping ensure they are deployed to serve citizens and meet national priorities in science, education, resilience, and public services.Building on our collaborations with the US and UK governments, we are establishing a new partnership with Indian government bodies and local institutions. In the global AI transformation, India is showing exceptional leadership in applying the technology to tackle its own biggest challenges. But India is going even further, playing a critical international role by convening this week the fourth global AI summit of governments, companies and civil society. International dialogue and collaboration will guide positive impacts and create the global frameworks required to prepare society for a future with AI.Partnership in India to broaden AI accessOur partnerships are designed to accelerate the pace of progress across India. Here are a few ways we are working together to unlock new possibilities in science and education.Advancing scientific breakthroughsGoogle DeepMind, Google Research and Google.org are partnering with the Anusandhan National Research Foundation (ANRF) to facilitate the adoption of AI models to advance science. We’re providing access to our frontier AI for Science models, supporting hackathons and community contests, and enabling training and mentorship to students, researchers, and those in the early stages of their careers.Researchers and engineers in India will be able to use our AI tools, including:AlphaGenome: An AI model to help scientists better understand how mutations in human DNA sequences impact a wide range of gene functionsAI Co-scientist: A multi-agent AI system that acts as a virtual scientific collaboratorEarth AI: A collection of models built on Gemini’s advanced reasoning that are helping enterprises, nonprofits, and cities with everything from environmental monitoring to disaster responseScientists around the world are already using AlphaFold – our AI system capable of accurately predicting the structure and interactions of proteins, DNA, RNA, ligands and more – to accelerate discoveries. India stands as the fourth largest adopter of AlphaFold globally, with over 180,000 researchers using it today. We hope to see Indian scientists benefit even more from using AlphaGenome and the other AI systems we are now providing.We’re also working to support AI for science at a global level. This is why, today at the India Summit, we announced the $30 million Google.org Impact Challenge: AI for Science, an open call for researchers, nonprofits, and social enterprises in India, and around the world, using AI to achieve scientific breakthroughs. Selected awardees will also have the opportunity to participate in a Google.org Accelerator, receiving engineering support, expert mentorship, and infrastructure from Google DeepMind and Google Research to turn their concepts into scalable discoveries.Empowering India’s Students and Teachers with an AI-powered FutureOur recent survey with Ipsos has shown that learning is the top motivation for using AI globally. This is especially true in India, which now leads the world in daily Gemini usage by students. We’re seeing AI can drive profound comprehension and critical thinking when it is purpose-built for learning and implemented as a supportive partner to educators.At City Montessori School in Lucknow, teachers are integrating Guided Learning into math classes for Grade 8-9 students and seeing a positive response. An early analysis of a randomized control study conducted by Fab AI shows that students are demonstrating a desire for deeper learning, not just quick answers: in almost three out of every four conversations on Gemini, students sought to develop their understanding rather than a quick answer or shortcut.That’s why we’re expanding efforts with additional partners to supercharge the potential of learning for more Indian students and teachers:Powering innovation hubs with GenAI assistants: Together with Atal Tinkering Labs, which serves more than 10,000 Indian schools and 11 million students, we will help incorporate robotics and coding into local curricula, integrate Gemini thoughtfully into teacher workflows, and build a safely guardrailed AI assistant for students grounded in national curriculum standards that can act as an educational partner. Teachers can access real-time tips to help students fix a robot missing a part with readily available materials or mend a broken circuit design by simply pointing a camera to it or asking Gemini in chat.Transforming textbooks into interactive digital journeys: In a first-of-its-kind partnership with PM Publishers Pvt. Ltd., a K-12 textbook publisher in India, Gemini will be used to transform two million static textbooks into AI-powered interactive journeys across more than 250 titles and 2,000 schools. Each book features a QR code that can be scanned by students to access a custom Gem (specialized versions of the Gemini AI model), that acts as an expert assistant on the subject, providing summaries and responses on the contents of the respective book.Serving India’s linguistic diversity: There is incredible potential for AI to make a positive impact on education when built in close partnership with experts and grounded in local language and culture. Building on Google.org’s recent $2 million founding contribution to establish the new Indic Language Technologies Research Hub at IIT Bombay, we’ll help incorporate India’s linguistic diversity into AI as it advances globally.These efforts build on the global success of existing AI literacy programs like Experience AI, a joint partnership developed by Google DeepMind with Raspberry Pi Foundation, which has already reached up to 300,000 students and 8,000 teachers in India.AI solutions for India’s agriculture and energy sectorsOur new partnerships in science and education build on our ongoing collaboration with local Indian organizations to tackle global challenges in agriculture and energy security. Working with Indian startups, institutions like Council on Energy, Environment and Water (CEEW), and Indian state and central government entities are using the APIs of our freely available Agri AI models to enhance agricultural resilience, crop productivity and farmer incomes. TerraStack is also using Google AI to combine satellite, crop, and weather data, into hyper-local insights that help farmers make better agricultural decisions.We also recently announced a growing collaboration with Open Climate Fix to integrate our WeatherNext AI models into India’s electricity grid operations. We’re aiming to significantly improve the accuracy of renewable energy forecasts in India, help grid operators manage volatility, and support the country’s ambitious clean energy targets. When we tested the integration of WeatherNext into OCF’s wind generation forecast, results showed up to 8% accuracy improvement in forecast performance.This partnership comes as India rapidly scales its renewable capacity, becoming the third largest generator of solar energy globally in 2023, with an ambitious target of installing 500 GW of renewable capacity by 2030. Working together on energy solutions has never been more important – we remain committed to working with experts in India to progress this effort together to prepare for the future.Preparing for the future togetherAI’s global impact is inevitable, but its success is not. To turn potential into prosperity, we are committing to deep, local collaboration with India’s government bodies and institutions to ensure AI delivers tangible results across the subcontinent–and the world.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. The curious case of the disappearing Lamborghinis Across the world, unsuspecting people are unwittingly becoming caught up in a new and growing type of organized criminal enterprise: vehicle transport fraud and theft.Crooks use email phishing, fraudulent paperwork, and other tactics to impersonate legitimate transport companies and get hired to deliver a luxury vehicle. They divert the shipment away from its intended destination before using a mix of technology, computer skills, and old-school techniques to erase traces of the vehicle’s original ownership and registration. In some cases, the car has been resold or is out of the country by the time the rightful owner even realizes it’s missing.The nationwide epidemic of vehicle transport fraud and theft has remained under the radar, even as it’s rocked the industry over the past two years. MIT Technology Review identified more than a dozen cases involving high-end vehicles, obtained court records, and spoke to law enforcement, brokers, drivers, and victims in multiple states to reveal how transport fraud is wreaking havoc across the country. Read the full story. —Craig Silverman
The scientist using AI to hunt for antibiotics just about everywhere
Antimicrobial resistance is a major problem. Infections caused by bacteria, fungi, and viruses that have evolved ways to evade treatments are now associated with more than 4 million deaths per year, and a recent analysis predicts that number could surge past 8 million by 2050. Bioengineer and computational biologist César de la Fuente has a plan. His team at the University of Pennsylvania is training AI tools to search genomes far and deep for peptides with antibiotic properties. His vision is to assemble those peptides—molecules made of up to 50 amino acids linked together—into various configurations, including some never seen in nature. The results, he hopes, could defend the body against microbes that withstand traditional treatments—and his quest has unearthed promising candidates in unexpected places. Read the full story. —Stephen Ornes These stories are both from the next print issue of MIT Technology Review magazine, which is all about crime. If you haven’t already, subscribe now to receive future issues once they land.  The must-reads I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology. 1 The Pentagon is close to cutting all business ties with AnthropicThe move would force anyone who wants to deal with the US military to cease working with Anthropic too. (Axios)+ Claude was used in the US raid to capture the former Venezuelan President. (WSJ $)+ Generative AI is learning to spy for the US military. (MIT Technology Review)

2 RFK Jr is setting his sights on baby formulaBut advocacy groups are concerned about how grounded in science the administration’s overhaul suggestions will be. (WSJ $)3 Germany is edging closer to banning social media for under-16sIn an effort to create safer digital spaces for young web users. (Bloomberg $)+ The country’s centre-left is in agreement with their conservative coalition partners. (Reuters) 4 Creative hackers are fighting back against ICEThe maker community is resisting through laser-cutting and 3D-printing. (Wired $)+ ICE has signed hundreds of deals with local law enforcement. (NBC News) 5 Consultancies have built thousands of AI agentsNow it’s time to see if they can actually deliver. (Insider $)+ Don’t let hype about AI agents get ahead of reality. (MIT Technology Review) 6 Restaurant workers are sick of being recorded 👓Meta’s smart glasses make the video-recording process more surreptitious than ever. (NYT $) 7 The Arctic’s rivers are turning bright orangeBut it’s climate change, not mining, that’s to blame. (FT $)+ What’s going to happen now the EPA can no longer fight climate change? (Undark)+ Scientists can see Earth’s permafrost thawing from space. (MIT Technology Review) 8 NASA let AI drive its Mars Perseverance roverIt traversed 456 meters across two days without human intervention. (IEEE Spectrum)+ That’s…not very fast at all. (Semafor)+ Slow-moving food delivery robots are under attack in the US. (Economist $)9 This machine is able to translate photos into smellsSelect your images very carefully, is my advice.(Fast Company $)10 One of YouTube’s biggest creators is now a successful director Mark Fischbach funded, made and released his film in theaters entirely independently. (The Atlantic $) Quote of the day
“My advice to them would be to get with the program.” —Jeremy Newmark, leader of a British council near the town of Potters Bar, has some choice words for the locals disputing plans to build a massive AI data center nearby, Wired reports.
One more thing The quest to find out how our bodies react to extreme temperaturesClimate change is subjecting vulnerable people to temperatures that push their limits. In 2023, about 47,000 heat-related deaths are believed to have occurred in Europe. Researchers estimate that climate change could add an extra 2.3 million European heat deaths this century. That’s heightened the stakes for solving the mystery of just what happens to bodies in extreme conditions.While we broadly know how people thermoregulate, the science of keeping warm or cool is mottled with blind spots. Researchers around the world are revising rules about when extremes veer from uncomfortable to deadly. Their findings change how we should think about the limits of hot and cold—and how to survive in a new world. Read the full story. —Max G.Levy We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or skeet ’em at me.) + I guarantee you’ve never seen a diner that looks quite like the Niemeyer Sphere.+ How New Yorkers keep partying in sub-zero temperatures.+ The interiors of Love Story, the new show chronicling the lives of John F. Kennedy Jr. and Carolyn Bessette, are a ‘90s dream.+ Ever wondered why some people see certain colors a certain way? Wonder no more.

When Sam Zahr first saw the gray Rolls-Royce Dawn convertible with orange interior and orange roof, he knew he’d found a perfect addition to his fleet. “It was very appealing to our clientele,” he told me. As the director of operations at Dream Luxury Rental, he outfits customers in the Detroit area looking to ride in style to a wedding, a graduation, or any other event with high-end vehicles—Rolls-Royces, Lamborghinis, Bentleys, Mercedes G-Wagons, and more. But before he could rent out the Rolls, Zahr needed to get the car to Detroit from Miami, where he bought it from a used-car dealer.  His team posted the convertible on Central Dispatch, an online marketplace that’s popular among car dealers, manufacturers, and owners who want to arrange vehicle shipments. It’s not too complicated, at least in theory: A typical listing includes the type of vehicle, zip codes of the origin and destination, dates for pickup and delivery, and the fee. Anyone with a Central Dispatch account can see the job, and an individual carrier or transport broker who wants it can call the number on the listing. Zahr’s team got a call from a transport company that wanted the job. They agreed on the price and scheduled pickup for January 17, 2025. Zahr watched from a few feet away as the car was loaded into an enclosed trailer. He expected the vehicle to arrive in Detroit just a few days later—by January 21.  But it never showed up. Zahr called a contact at the transport company to ask what happened. 
“He’s like, I don’t know what you’re talking about.”  Zahr told me his contact angrily told him they mostly ship Coca-Cola products, not luxury cars. “He was yelling and screaming about it,” Zahr said.
Over the years, people have broken into his business to steal cars, or they’ve rented them out and never come back. But until this day, he’d never had a car simply disappear during shipping. He’d expected no trouble this time around, especially since he’d used Central Dispatch—“a legit platform that everyone uses to transport cars,” he said.  “That’s the scary part about it, you know?” Wreaking havoc Zahr had unwittingly been caught up in a new and growing type of organized criminal enterprise: vehicle transport fraud and theft. Crooks use email phishing, fraudulent paperwork, and other tactics to impersonate legitimate transport companies and get hired to deliver a luxury vehicle. They divert the shipment away from its intended destination and then use a mix of technology, computer skills, and old-school chop shop techniques to erase traces of the vehicle’s original ownership and registration. These vehicles can be retitled and resold in the US or loaded into a shipping container and sent to an overseas buyer. In some cases, the car has been resold or is out of the country by the time the rightful owner even realizes it’s missing. “Criminals have learned that stealing cars via the web portals has become extremely easy, and when I say easy—it’s become seamless,” says Steven Yariv, the CEO of Dealers Choice Auto Transport of West Palm Beach, Florida, one of the country’s largest luxury-vehicle transport brokers. Individual cases have received media coverage thanks to the high value of the stolen cars and the fact that some belong to professional athletes and other celebrities. In late 2024, a Lamborghini Huracán belonging to Colorado Rockies third baseman Kris Bryant went missing en route to his home in Las Vegas; R&B singer Ray J told TMZ the same year that two Mercedes Maybachs never arrived in New York as planned; and last fall, NBA Hall of Famer Shaquille O’Neal had a $180,000 custom Range Rover stolen when the transport company hired to move the vehicle was hacked. “They’re saying they think it’s probably in Dubai by now, to be honest,” an employee of the company that customized the SUV told Shaq in a YouTube video.
“Criminals have learned that stealing cars via the web portals has become extremely easy, and when I say easy—it’s become seamless.” Steven Yariv, CEO, Dealers Choice Auto Transport of West Palm Beach, Florida But the nationwide epidemic of vehicle transport fraud and theft has remained under the radar, even as it’s rocked the industry over the past two years. MIT Technology Review identified more than a dozen cases involving high-end vehicles, obtained court records, and spoke to law enforcement, brokers, drivers, and victims in multiple states to reveal how transport fraud is wreaking havoc across the country. RICHARD CHANCE It’s challenging to quantify the scale of this type of crime, since there isn’t a single entity or association that tracks it. Still, these law enforcement officials and brokers, as well as the country’s biggest online car-transport marketplaces, acknowledge that fraud and theft are on the rise.  When I spoke with him in August, Yariv estimated that around 8,000 exotic and high-end cars had been stolen since the spring of 2024, resulting in over $1 billion in losses. “You’re talking 30 cars a day [on] average is gone,” he said. Multiple state and local law enforcement officials told MIT Technology Review that the number is plausible. (The FBI did not respond to a request for an interview.)
“It doesn’t surprise me,” said J.D. Decker, chief of the Nevada Department of Motor Vehicles’ police division and chair of the fraud subcommittee for the American Association of Motor Vehicle Administrators. “It’s a huge business.” Data from the National Insurance Crime Bureau (NICB), a nonprofit that works with law enforcement and the insurance industry to investigate insurance fraud and related crimes, provides further evidence of this crime wave. NICB tracks both car theft and cargo theft, a broad category that refers to goods, money, or baggage that is stolen while part of a commercial shipment; the category also covers cases in which a vehicle is stolen via a diverted transport truck or a purloined car is loaded into a shipping container. NICB’s statistics about car theft show that it has declined following an increase during the pandemic—but over the same period cargo theft has dramatically increased, to an estimated $35 billion annually. The group projected in June that it was expected to rise 22% in 2025. NICB doesn’t break out data for vehicles as opposed to other types of stolen cargo. But Bill Woolf, a regional director for the organization, said an antifraud initiative at the Port of Baltimore experienced a 200% increase from 2023 to 2024 in the number of stolen vehicles recovered. He said the jump could be due to the increased effort to identify stolen cars moving through the port, but he noted that earlier the day we spoke, agents had recovered two high-end stolen vehicles bound for overseas. “One day, one container—a million dollars,” he said.
Many other vehicles are never recovered—perhaps a result of the speed with which they’re shipped off or sold. Travis Payne, an exotic-car dealer in Atlanta, told me that transport thieves often have buyers lined up before they take a car: “When they steal them, they have a plan.”  In 2024, Payne spent months trying to locate a Rolls-Royce he’d purchased after it was stolen via transport fraud. It eventually turned up in the Instagram feed of a Mexican pop star, he says. He never got the car back. The criminals are “gonna keep doing it,” he says, “because they make a couple phone calls, make a couple email accounts, and they get a $400,000 car for free. I mean, it makes them God, you know?” Out-innovating the industry The explosion of vehicle transport fraud follows a pattern that has played out across the economy over the past roughly two decades: A business that once ran on phones, faxes, and personal relationships shifted to online marketplaces that increased efficiency and brought down costs—but the reduction in human-to-human interaction introduced security vulnerabilities that allowed organized and often international fraudsters to enter the industry. In the case of vehicle transport, the marketplaces are online “load boards” where car owners, dealerships, and manufacturers post about vehicles that need to be shipped from one location to another. Central Dispatch claims to be the largest vehicle load board and says on its website that thousands of vehicles are posted on its platform each day. It’s part of Cox Automotive, an industry juggernaut that owns major vehicle auctions, Autotrader, Kelley Blue Book, and other businesses that work with auto dealers, lenders, and buyers. The system worked pretty well until roughly two years ago, when organized fraud rings began compromising broker and carrier accounts and exploiting loopholes in government licensing to steal loads with surprising ease and alarming frequency.
A theft can start with a phishing email that appears to come from a legitimate load board. The recipient, a broker or carrier, clicks a link in the message, which appears to go to the real site—but logging in sends the victim’s username and password to a criminal. The crook logs in as the victim, changes the account’s email and phone number to reroute all communications, and begins claiming loads of high-end vehicles. Cox Automotive declined an interview request but said in a statement that the “load board system still works well” and that “fraud impacts a very small portion” of listings. “Every time we come up with a security measure to prevent the fraudster, they come up with a countermeasure.” Bill Woolf, a regional director, National Insurance Crime Bureau Criminals also gain access to online marketplaces by exploiting a lax regulatory environment. While a valid US Department of Transportation registration is required to access online marketplaces, it’s not hard for bad actors to register sham transport companies and obtain a USDOT number from the Federal Motor Carrier Safety Administration, the agency that regulates commercial motor vehicles. In other cases, criminals compromise the FMCSA accounts of legitimate companies and change their phone numbers and email addresses in order to impersonate them and steal loads. (USDOT did not respond to a request for comment.)
As Bek Abdullayev, the founder of Super Dispatch, one of Central Dispatch’s biggest competitors, explained in an episode of the podcast Auto Transport Co-Pilot, “FMCSA [is] authorizing people that are fraudulent companies—people that are not who they say they are.” He added that people can “game the system and … obtain paperwork that makes [them] look like a legitimate company.” For example, vehicle carrier insurance can be obtained quickly—if temporarily—by submitting an online application with fraudulent payment credentials. The bottom line is that crooks have found myriad ways to present themselves as genuine and permitted vehicle transport brokers and carriers. Once hired to move a vehicle, they often repost the car on a load board using a different fraudulent or compromised account. While this kind of subcontracting, known as “double-­brokering,” is sometimes used by companies to save money, it can also be used by criminals to hire an unwitting accomplice to deliver the stolen car to their desired location. “They’re booking cars and then they’re just reposting them and dispatching them out to different routes,” says Yariv, the West Palm Beach transport broker.  “A lot of this is cartel operated,” says Decker, of the Nevada DMV, who also serves on a vehicle fraud committee for the International Association of Chiefs of Police. “There’s so much money in it that it rivals selling drugs.” Even though this problem is becoming increasingly well known, fraudsters continue to steal, largely with impunity. Brokers, auto industry insiders, and law enforcement told MIT Technology Review that load boards and the USDOT have been too slow to catch and ban bad actors. (In its statement, Cox Automotive said it has been “dedicated to continually enhancing our processes, technology, and education efforts across the industry to fight fraud.”) Jake MacDonald, who leads Super Dispatch’s fraud monitoring and investigation efforts, put it bluntly on the podcast with Abdullayev: the reason that fraud is “jumping so much” is that “the industry is slowly moving over to a more technologically advanced position, but it’s so slow that fraud is actually [out-]innovating the industry.” A Florida sting As it turns out, the person Zahr’s team hired on Central Dispatch didn’t really work for the transport company.  After securing the job, the fraudster reposted the orange-and-gray Rolls convertible to a load board. And instead of saying that the car needed to go from Miami to the real destination of Detroit, the new job listed an end point of Hallandale Beach, Florida, just 20 or so miles away. It was a classic case of malicious double-­brokering: the crooks claimed a load and then reposted it in order to find a new, unsuspecting driver to deliver the car into their possession. On January 17 of last year, the legitimate driver showed up in a Dodge Ram and loaded the Rolls into an enclosed trailer as Zahr watched.
“The guy came in and looked very professional, and we took a video of him loading the car, taking pictures of everything,” Zahr told me. He never thought to double-­check where the driver was headed or which company he worked for. Not long after a panicked Zahr spoke with his contact at the transport company he thought he was working with, he reported the car as stolen to the Miami police. Detective Ryan Chin was assigned to the case. It fit with a pattern of high-end auto theft that he and his colleagues had recently been tracking. “Over the past few weeks, detectives have been made aware of a new method on the rise for vehicles being stolen by utilizing Central Dispatch,” Chin wrote in records obtained by MIT Technology Review. “Specific brokers are re-routing the truck drivers upon them picking up vehicles posted for transport and routing them to other locations provided by the broker.”  Chin used Zahr’s photos and video to identify the truck and driver who’d taken the Rolls. By the time police found him, on January 31, the driver had already dropped off Zahr’s Rolls in Hallandale Beach. He’d also picked up and delivered a black Lamborghini Urus and a White Audi R8 for the same client. Each car had been stolen via double-brokering transport fraud, according to court records.  The police department declined to comment or to make Chin available for an interview. But a source with knowledge of the case said the driver was “super cooperative.” (The source asked not to be identified because they were not authorized to speak to the media, and the driver does not appear to have been identified in court records.) The driver told police that he had another load to pick up at a dealership in Naples, Florida, later that same day—a second Lamborghini Urus, this one orange. Police later discovered it was supposed to be shipped to California. But the carrier had been hired to bring the car, which retails for about $250,000, to a mall in nearby Aventura. He told police that he suspected it was going to be delivered to the same person who had booked him for the earlier Rolls, Audi, and Lamborghini deliveries, since “the voice sounds consistent with who [the driver] dealt with prior on the phone.” This drop-off was slated for 4 p.m. at the Waterways Shoppes mall in Aventura. That was when Chin and a fellow detective, Orlando Rodriguez, decided to set up a sting.  The officers and colleagues across three law enforcement agencies quickly positioned themselves in the Waterways parking lot ahead of the scheduled delivery of the Urus. They watched as, pretty much right on schedule that afternoon, the cooperative driver of the Dodge Ram rolled to a stop in the palm-tree-lined lot, which was surrounded by a kosher supermarket, Japanese and Middle Eastern restaurants, and a physiotherapy clinic. The driver went inside the trailer and emerged in the orange Lamborghini. He parked it and waited near the vehicle. Roughly 30 minutes later, a green Rolls-Royce Cullinan (price: $400,000 and up) arrived with two men and a teenager inside. They got out, opened the trunk, and sat on the tailgate of the vehicle as one man counted cash. “They’re doing countersurveillance, looking around,” the source told me later. “It’s a little out of the ordinary, you know. They kept being fixated [on] where the truck was parked.”  The transport driver and the three males who arrived in the Rolls-Royce did not interact. But soon enough, another luxury vehicle, a Bentley Continental GT, which last year retailed for about $250,000 and up, pulled in. The Bentley driver got out, took the cash from one of the men sitting on the back of the Rolls, and walked over to the transport driver. He handed him $700 and took the keys to the Lamborghini. That’s when more than a dozen officers swooped in. “They had nowhere to go,” the source told me. “We surrounded them.” The two men in the Rolls were later identified as Arman Gevorgyan and Hrant Nazarian, and the man in the Bentley as Yuriy Korotovskyy. The three were arrested and charged with dealing in stolen property, grand theft over $100,000, and organized fraud. (The teenager who arrived in the Rolls was Gevorgyan’s son. He was detained and released, according to Richard Cooper, Gevorgyan’s attorney.) As investigators dug into the case, the evidence suggested that this was part of the criminal pattern they’d been following. “I think it’s organized,” the source told me. It’s something that transport industry insiders have talked about for a while, according to Fred Mills, the owner of Florida-based Advantage Auto Transport, a company that specializes in transporting high-end vehicles. He said there’s even a slang term to describe people engaged in transport fraud: the flip-flop mafia. 
.cst-large,
.cst-default {
width: 100%;
}

@media (max-width: 767px) {
.cst-block {
overflow-x: hidden;
}
}

@media (min-width: 630px) {
.cst-large {
margin-left: -25%;
width: 150%;
}

@media (min-width: 960px) {
.cst-large {
margin-left: -16.666666666666664%;
width: 140.26%;
}
}

@media (min-width: 1312px) {
.cst-large {
width: 145.13%;
}
}
} @media (min-width: 60rem) { .flourish-embed { width: 60vw; transform: translateX(-50%); left: 50%; position: relative; } } It has multiple meanings. One is that the people who show up to transport or accept a vehicle “are out there wearing, you know, flip-flops and slides,” Mills says. The second refers to how fraudsters “flip” from one carrier registration to another as they try to stay ahead of regulators and complaints. In addition to needing a USDOT number, carriers working across states need an interstate operating authority (commonly known as an MC number) from the USDOT. Both IDs are typically printed on the driver and passenger doors. But the rise of ­double-brokering—and of fly-by-night and fraudulent carriers—means that drivers increasingly just tape IDs to their door.  Mills says fraudsters will use a USDOT number for 10 or 11 months, racking up violations, and then tape up a new one. “They just wash, rinse, and repeat,” he says. Decker from the Nevada DMV says a lot of high-end vehicles are stolen because dealerships and individual customers don’t properly check the paperwork or identity of the person who shows up to transport them. “‘Flip-flop mafia’ is an apt nickname because it’s surprisingly easy to get a car on a truck and convince somebody that they’re a legitimate transport operation when they’re not,” he says. Roughly a month after it disappeared, Zahr’s Rolls-Royce was recovered by the Miami Beach Police. Video footage obtained by a local TV station showed the gray car with its distinctive orange top being towed into a police garage.  What happens in Vegas Among the items confiscated from the men in Florida were $10,796 in cash and a GPS jammer. Law enforcement sources say jammers have become a core piece of technology for modern car thieves—necessary to disable the location tracking provided by GPS navigation systems in most cars. “Once they get the vehicles, they usually park them somewhere [and] put a signal jammer in there or cut out the GPS,” the Florida source told me. This buys them time to swap and reprogram the vehicle identification number (VIN), wipe car computers, and reprogram fobs to remove traces of the car’s provenance.  No two VINs are the same, and each is assigned to a specific vehicle by the manufacturer. Where they’re placed inside a vehicle varies by make and model. The NICB’s Woolf says cars also have confidential VINs located in places—including their electronic components—that are supposed to be known only to law enforcement and his organization. But criminals have figured out how to find and change them. “It’s making it more and more difficult for us to identify vehicles as stolen,” Woolf says. “Every time we come up with a security measure to prevent the fraudster, they come up with a countermeasure.” All this doesn’t even take very much time. “If you know what you’re doing, and you steal the car at one o’clock today, you can have it completely done at two o’clock today,” says Woolf. A vehicle can be rerouted, reprogrammed, re-VINed, and sometimes even retitled before an owner files a police report. That appears to have been the plan in the case of the stolen light-gray 2023 Lamborghini Huracán owned by the Rockies’ Kris Bryant. On September 29, 2024, a carrier hired via a load board arrived at Bryant’s home in Cherry Hills, Colorado, to pick up the car. It was supposed to be transported to Bryant’s Las Vegas residence within a few days. It never showed up there—but it was in fact in Vegas. Using Flock traffic cameras, which capture license plate information in areas across the country, Detective Justin Smith of the Cherry Hills Village Police Department tracked the truck and trailer that had picked up the Lambo to Nevada, and he alerted local police. On October 7, a Las Vegas officer spotted a car matching the Lamborghini’s description and pulled it over. The driver said the Huracán had been brought to his auto shop by a man whom the police were able to identify as Dat Viet Tieu. They arrested Tieu later that same day. In an interview with police, he identified himself as a car broker. He said he was going to resell the Lamborghini and that he had no idea that the car was stolen, according to the arrest report.  Police searched a Jeep Wrangler that Tieu had parked nearby and discovered it had been stolen—and had been re-VINed, retitled, and registered to his wife. Inside the car, police discovered “multiple fraudulent VIN stickers, key fobs to other high-end stolen vehicles, and fictitious placards,” their report said.  One of the fake VINs matched the make and model of Bryant’s Lamborghini. (Representatives for Bryant and the Rockies did not respond to a request for comment.)  Tieu was released on bail. But after he returned to LVPD headquarters two days later, on October 9, to reclaim his personal property, officers secretly placed him under surveillance with the hope that he’d lead them to one of the other stolen cars matching the key fobs they’d found in the Jeep.  It didn’t take long for them to get lucky. A few hours after leaving the police station, Tieu drove to Harry Reid International Airport, where he picked up an unidentified man. They drove to the Caesars Palace parking garage and pulled in near a GMC Sierra. Over the next three hours, the man worked on a laptop inside and outside the vehicle, according to a police report. At one point, he and Tieu connected jumper cables from Tieu’s rented Toyota Camry to the Sierra. “At 2323 hours, the white male adult enters the GMC Sierra, and the vehicle’s ignition starts. It was readily apparent the [two men] had successfully re-programmed a key fob to the GMC Sierra,” the report said. An officer watched as the man gave two key fobs to Tieu, who handed the man an unknown amount of cash. Still, the police let the men leave the garage.  The police kept Tieu and his wife under surveillance for more than a week. Then, on October 18, fearing the couple was about to leave town, officers entered Nora’s Italian Restaurant just off the Vegas Strip and took them into custody. “Obviously, we meet again,” a detective told Tieu. “I’m not surprised,” Tieu replied.  Police later searched the VIN on the Sierra from the Caesars lot and found that it had been reported stolen in Tremonton, Utah, roughly two weeks earlier. They eventually returned both the Sierra and Kris Bryant’s Lamborghini to their owners.  Tieu pleaded guilty to two felony counts of possession of a stolen vehicle and one count of defacing, altering, substituting, or removing a VIN. In October, he was sentenced to up to one year of probation; if it’s completed successfully, the plea agreement says, the counts of possession of a stolen vehicle will be dismissed. His attorneys, David Z. Chesnoff and Richard A. Schonfeld, said in a statement that they were “pleased” with the court’s decision, “in light of [Tieu’s] acceptance of responsibility.”  Taking the heat Many vehicles stolen via transport fraud are never recovered. Experts say the best way to stop this criminal cycle would be to disrupt it before it starts.  That would require significant changes to the way that load boards operate. Bryant’s Lamborghini, Zahr’s and Payne’s Rolls-Royces, and the orange Lamborghini Urus in Florida were all posted for transport on Central Dispatch. Both brokers and shippers argue that the company hasn’t taken enough responsibility for what they characterize as weak oversight. “If the crap hits the fan, it’s on us as a broker, or it’s on the trucking company … they have no liability in the whole transaction process. So it definitely frosted a lot of people’s feathers.” Fred Mills, owner of Florida-based Advantage Auto Transport “You’re Cox Automotive—you’re the biggest car company in the world for dealers—and you’re not doing better screenings when you sign people up?” says Payne. (The spokesperson for Cox Automotive said that it has “a robust verification process for all clients … who sign up.”) “If the crap hits the fan, it’s on us as a broker, or it’s on the trucking company, or the clients’ insurance, [which means] that they have no liability in the whole transaction process,” says Mills. “So it definitely frosted a lot of people’s feathers.” Over the last year, Central Dispatch has made changes to further secure its platform. It introduced two-factor authentication for user accounts and started enabling shippers to use its app to track loads in real time, among other measures. It also kicked off an awareness campaign that includes online educational content and media appearances to communicate that the company takes its responsibilities seriously. “We’ve removed over 500 accounts already in 2025, and we’ll continue to take any of that aggressive action where it’s needed,” said Lainey Sibble, Central Dispatch’s head of business, in a sponsored episode of the Auto Remarketing Podcast. “We also recognize this is not going to happen in a silo. Everyone has a role to play here, and it’s really going to take us all working together in partnership to combat this issue.” Mills says Central Dispatch got faster at shutting down fraudulent accounts toward the end of last year. But it’s going to take time to fix the industry, he adds: “I compare it to a 15-year opioid addiction. It’s going to take a while to detox the system.”  Yariv, the broker in West Palm Beach, says he has stopped using Central Dispatch and other load boards altogether. “One person has access here, and that’s me. I don’t even log in,” he told me. His team has gone back to working the phones, as evidenced by the din of voices in the background as we spoke.  RICHARD CHANCE “[The fraud is] everywhere. It’s constant,” he said. “The only way it goes away is the dispatch boards have to be shut down—and that’ll never happen.” It also remains to be seen what kind of accountability there will be for the alleged thieves in Florida. Korotovskyy and Nazarian pleaded not guilty; as of press time, their trials were scheduled to begin in May. (Korotovskyy’s lawyer, Bruce Prober, said in a statement that the case “is an ongoing matter” and his client is “presumed innocent,” while Nazarian’s attorney, Yale Sanford, said in a statement, “As the investigation continues, Mr. Nazarian firmly asserts his innocence.” A spokesperson with Florida’s Office of the State Attorney emailed a statement: “The circumstances related to these arrests are still a matter of investigation and prosecution. It would be inappropriate to be commenting further.”) In contrast, Gevorgyan, the third man arrested in the Florida sting, pleaded guilty to four charges.  Yet he maintains his innocence, according to Cooper, his lawyer: “He was pleading [guilty] to get out and go home.” Cooper describes his client as a wealthy Armenian national who runs a jewelry business back home, adding that he was deported to Armenia in September.  Cooper says his client’s “sweetheart” plea deal doesn’t require him to testify or otherwise supply information against his alleged co-conspirators—or to reveal details about how all these luxury cars were mysteriously disappearing across South Florida. Cooper also says prosecutors may have a difficult time convicting the other two men, arguing that police acted prematurely by arresting the trio without first seeing what, if anything, they intended to do with the Lamborghini. “All they ever had,” Cooper says, “was three schmucks sitting outside of the Lamborghini.”  Craig Silverman is an award-winning journalist and the cofounder of Indicator, a publication that reports on digital deception.