Yahoo News (Page 36)

The phrase “touch grass” has become the internet’s way of telling someone to log off and rejoin the real world. Erik Torenberg, a general partner at Andreessen Horowitz, thinks the phrase has it exactly backward — and that getting the philosophy right has enormous economic consequences.

In a new essay published through a16z, Torenberg makes a sweeping argument: the internet isn’t encroaching on real life. It has become real life. And what looks like a cultural provocation is, on closer reading, a business thesis about where value will be created in an economy being remade by artificial intelligence.

“The internet is real life,” Torenberg writes. “And navigating life means navigating the internet.”

Upstream of everything

The evidence Torenberg marshals ranges across culture, politics, language, and media. News now “exists to summarize things that have already happened online.” Music is being restructured by TikTok’s 15-second clip format, the way radio once defined the verse-chorus arrangement. Politicians are fluent in meme-speak — J.D. Vance discouraging “blackpilling” — because their staffers and constituencies are shaped by internet discourse. Even language no longer merely spreads through the internet: it originates there.

The deeper claim is philosophical. Torenberg argues there is no such thing as an unmediated human existence — and never was. “From the beginning of history, we’ve used technology to mediate between ourselves and the world,” he writes. Domesticating horses, inventing currency, building governments — each was a mediating layer between humanity and raw nature. The internet is simply the newest and most expansive version of that ancient process, humans learning to interface with technology. “Even real life is not ‘real life.’”

In a follow-up email to Fortune, Torenberg refined the philosophical claim. Mediating attention and perception, he noted, is not unique to the internet — governments, currency, organized religion, and even horses all did that in their own ways. “If you believe that language is partly a technology, rather than an entirely in-built aspect of the human animal,” he added, “then it would certainly count as well.” What makes the internet distinct isn’t the fact of mediation but its scale and what he called its “bespokeness” — the degree to which a person can lose themselves in a fully personalized experience. It is that combination, he argued, that makes the online/offline dichotomy the real illusion, not the internet’s claim to be real life.

On the question of whether humanity has reached any equilibrium with this new mediating layer, Torenberg was direct: we haven’t, and may not for some time. He pointed to two forces likely to shape the eventual settlement. The first is cultural: norms around online speech are still very much in flux. He cited figures like Vice President JD Vance, calling for greater tolerance of aggressive online rhetoric, on the grounds that humans evolved to speak without assuming that their words would be permanently, verbatim recorded for the world to see. The second force is biological. “Some subset of people will get ‘oneshotted’ by technology in a way that prevents them from reproducing,” Torenberg said — citing, as an extreme example, mobile gaming addiction displacing human interaction entirely. He pointed to declining global fertility numbers as one possible signal, suggesting that many people simply find the possibilities the internet has opened up more compelling than starting a family, and that “we’ll see strong selection against whatever genes influence those preferences before we reach equilibrium.” His historical analogy was pointed: “That’s the story of humanity. The Yamnaya’s domestication of the horse didn’t go great for Early European Farmers.”

A historical echo

It is a thesis that finds an unlikely illustration in a separate essay published the same week by George Mason economist Alex Tabarrok. Writing on his blog Marginal Revolution, Tabarrok makes the increasingly familiar argument for the AI age that the Luddites — famous for smashing looms in early 19th-century England — were, in a sense, the first people to attack AI. But unlike most, he links the loom to its unlikely descendant: the computer.

The Jacquard loom, introduced in France around 1805, used a chain of punched cards to control weaving patterns, a design that Charles Babbage borrowed directly for his Analytical Engine and that eventually traced a line to the modern computer. He quotes from Ada Lovelace, the daughter of Lord Byron and, many think, the world’s first computer programmer, roughly 100 years before computers existed: “The Analytical Engine weaves algebraical patterns just as the Jacquard-loom weaves flowers and leaves.”

Tabarrok thanked Anthropic’s Claude for assistance in pulling his post on the Luddites together, and he clarified to Fortune that he was familiar with the link between the loom and Babbage’s Analytical Engine, but Claude helped him connect more dots: Manchester, the epicenter of both the Industrial Revolution and many Luddite riots, was also home of the Manchester Mark 1, the first electronic stored-program computer, where Alan Turing, father of modern computing, was hired to program it.

The loom is, in other words, a perfect illustration of Torenberg’s mediating-layer argument. It didn’t replace the weaver’s embodied existence — it inserted itself between the weaver’s skill and the finished cloth, restructuring what “weaving” meant and who could do it. Tabarrok argues that “programmable looms brought patterned clothes to the masses, surely a good thing in the long run, economically speaking, but surely also with some short-term pain during the transition to the new interface. Extending this to Torenberg’s argument, the internet has done the same thing to nearly every domain of human activity, at incomparably greater scale.

To be sure

Not everyone will accept the leap from “the internet shapes everything” to “the internet is real life.” Critics would note that Torenberg conflates influence with identity: a hammer shapes a house without being the house. Embodied experience — grief, illness, hunger, the irreducible fact of a body — still refuses to fully migrate online. The danger in collapsing the distinction is that decisions get made based on what is loud and visible in a feed rather than what is true in aggregate human experience.

Torenberg anticipates the objection, and his response is pointed: even telling someone to “touch grass” is itself internet-native language. The critics, he argues, have already proven his point: “When someone tells you that you are ‘extremely online,’ or need to ‘touch grass,’ they are–intentionally or not–confessing that they too have had their brain colonized by internet cliches.”

Where, what, and who

What makes the essay more than a cultural argument is the economic framework it implies — one that maps onto three questions economists are urgently asking about the AI economy.

Where is the new economy organized? Torenberg’s answer is unambiguous: the internet is now the primary mediating layer through which all experience, culture, and meaning flows. The business that helps people navigate that layer becomes critical infrastructure. That is the explicit bet behind Monitoring the Situation, the live online news channel a16z is backing as a direct extension of Torenberg’s thesis.

Torenberg was careful to note, however, that MTS is formally a separate entity — a16z is a minority investor alongside several other individuals and organizations. He pushed back on any suggestion of inherent conflict between being an investor in part of that mediating layer and analyzing it honestly. “The only way for MTS to be successful at what it wants to do is for it to be an open and honest channel of information,” he told Fortune. “The information is all out there already, so it’s not as if other people can’t take the idea and run with it.” The theoretical tension, he argued, would be less between investing and philosophizing than between investing and sharing one’s true insights publicly — either because of competitive risk or the temptation to talk one’s book. He believes transparency serves the ecosystem better.

On the pace of change in the space he’s analyzing, Torenberg acknowledged the challenge directly: even a comprehensive understanding of the internet today would not stay fresh for long. But he argued that uncertainty itself is a subject. He pointed to the Iran War as an example — mass use of AI, among other factors, has shaped public understanding of that conflict in ways that differ sharply from how the Russia-Ukraine conflict was processed four years earlier. “Huge numbers of people becoming convinced that Bibi Netanyahu was killed and that the Israeli government is producing deepfakes of him — that’s a story,” he said. “The uncertainty is a story.” He invoked Baudrillard’s famous provocation that the Gulf War “did not take place.” “Maybe the Iran War didn’t take place either,” Torenberg said. “But people still read Baudrillard’s essay.”

What becomes scarce within that layer? University of Chicago behavioral economist Alex Imas has made the complementary argument: as AI commoditizes information, content, and cognitive labor, what becomes economically valuable is the relational layer — the things with an irreducibly human element. His “relational sector” thesis holds that tomorrow’s middle-class consumption patterns will resemble those of the wealthy today, with people paying for human connection the way only the rich currently do. As he told Fortune recently, “There’s a lot of jobs right now that have a relational component, which will become relational jobs.”

This is Torenberg’s cultural argument translated directly into labor economics: if AI is commoditizing everything automatable within the internet’s mediating layer, then what’s scarce is authentic human navigation of that layer — precisely what Torenberg’s media network is selling.

Who captures the gains? This is where Tabarrok’s Luddite analogy cuts. The Luddites lost, he writes, not simply because programmable looms were better, but because the British military violently suppressed them and Parliament made frame-breaking a capital crime. As Tabarrok has separately noted, real British wages were flat from 1780 to 1840 while output per worker doubled; life expectancy in 1840s Manchester was 26. The gains finally broadened after 1840, and not through the market — they came through the Factory Acts, unions, and the hard construction of countervailing political power. As one commenter on Tabarrok’s post put it: “The gains were real. The distribution of those gains was not inevitable — it was enforced.”

“The first thing that people think about when they think about reducing work is unemployment,” Alex Tabarrok recently told Fortune. “But reducing work could mean, you know, a shorter work week. It could mean a longer retirement, a longer childhood, more holidays.”

That is the question Torenberg’s essay, by design, leaves unanswered. Torenberg identifies where the new economy is organized. Imas identifies what becomes valuable within it. Tabarrok’s history identifies who decides — and warns that the answer has never been determined by markets alone. If the internet is real life, and a16z holds significant infrastructure around how the internet-as-real-life is understood, the distribution question becomes pointed in ways that no amount of philosophical elegance can dissolve.

Torenberg did not respond to a request for comment.

For this story, Fortune journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.

This story was originally featured on Fortune.com

This post was originally published here

Palantir’s fourth-quarter earnings call turned into a geopolitical broadside as CEO Alex Karp blasted Canada and much of Europe for falling behind in the artificial intelligence race, casting the global economy as a looming conflict between “AI haves” and “have-nots.”​

Speaking after Palantir reported 70% year-over-year revenue growth to $1.407 billion in the fourth quarter and a rule of 40 score of 127, Karp argued that the company’s performance exposed a widening gap between countries and institutions willing to overhaul themselves around advanced AI software and those content to tinker at the margins.

Noting that Palantir’s U.S. business grew 93% year over year in the fourth quarter, with America now accounting for 77% of total revenue, Karp asked hypothetically: “What do bombastic numbers like this mean?” It’s actually bad news that Palantir is “doing things unlike any other company has done,” he argued, because it raises another question: “This obviously has import for the world. And what does it mean for the world?”

Karp as Davos Man 2.0

Echoing rhetoric from the Trump administration on display at the recent World Economic Forum meeting in Davos (where Karp was a speaker), the Palantir CEO offered a withering critique of the companies failing to adopt AI. “We’ve also seen, unfortunately, that there’s a real hesitance to adopt these kind of products in the West outside of America, and the two places leading here are China and America,” he said. “What we’re seeing in America is so widely divergent. And so the non‑adopters, the have‑nots, are hoping for a catch‑up function.” Good luck, he seemed to say, asserting that Palantir’s earnings are a “breakout function” that mean “the way in which we view value is obviously no longer relevant.”

The value being created by Palantir is “so large and so disproportionate that you can create a company that seemingly is exploding in terms of growth and quality of growth.” Then he named names, saying that Palantir sees adoption, sometimes wide-scale, of advanced AI platforms in parts of the Middle East and in China, but “lack of adoption in Canada, Northern Europe, and in Europe in general.” Just look at France, he said, one of the countries with “the clearest idea of the problem.” France has no alternative to solving this adoption problem and has been forced to keep signing new deals with Palantir. In December 2025, to that point, France renewed a three-year contract with the French intelligence services.

“One of the things you’re gonna see in Northern Europe, Canada, and other places is a real pressure to move to the left and right politically, very far,” Karp said. “Because the way you deal with this when you don’t have an answer to a question, you come up with ideologies that make no sense, and you try to implement them.”

To be sure, Karp’s framing ignores that Palantir itself has chosen to concentrate capacity on the U.S. and “doesn’t have the bandwidth” for more complex international work. It also dismisses legitimate reasons for slower or more selective adoption: European and Canadian regulatory regimes place a higher weight on privacy, civil liberties, and vendor diversity, with many governments preferring sovereign or domestic solutions in critical infrastructure. It also treats Palantir’s success in a uniquely favorable U.S.-defense‑centric market as if it were universal proof that countries like Canada and those in Europe are failing on AI simply because they are not buying his platform at scale. Different jurisdictions are entitled to pursue AI on their own timelines, with their own safeguards and mixes of vendors.

Analysts on Wall Street, as they are prone to do with such a hot stock, sided with Karp’s version of events. Bank of America Global Research, for instance, argued that Palantir’s blowout earnings constitute a “warning to slow adapters” on AI: “The clock is ticking.” Exponential growth is on display here following Palantir’s intentional actions on how to go to market, develop products, and be an enabler of AI decision-making, BofA wrote. If companies really want to be “AI companies,” analysts added, they need to provide real results. Allowing that the market’s relationship with AI companies “continues to be volatile,” BofA sees this set of results cementing Palantir’s place “as one which will survive and thrive in the chaos.”

The haves and have-nots

Inside companies, Karp and Palantir CTO Shyam Sankar described a similar split between AI “haves” and “have‑nots.” Chief revenue officer Ryan Taylor said some customers are now signing initial deals of $80 million to $96 million within months and rapidly expanding usage, citing examples of utility and energy clients whose annual contract values quadrupled or quintupled in 2025. Taylor framed those customers as “AI‑native enterprises” that start with large commitments and quickly scale to thousands of users and hundreds of use cases.​

“Our customers aren’t tentatively trying AI; they’re committing to it at scale,” Taylor said, adding that Palantir’s top 20 customers now generate an average of $94 million each in trailing 12‑month revenue, up 45% year over year. Karp argued that these firms are “defining the future of their industries,” while those still dabbling in pilots—the “AI have‑nots”—are “fighting for survival in the present.”

BofA noted how embedded Palantir is becoming in the corporate space, with an ever-expanding list of mentions in earnings calls, with 17 unique mentions this quarter, up from seven a year ago, and a new high of 38 total mentions, up from 25 in the year ago quarter.

Karp’s remarks came as Palantir leaned heavily into its role as a key supplier of AI-enabled systems to the U.S. government and defense sector. The company highlighted a U.S. Navy contract worth up to $448 million to modernize the shipbuilding supply chain and described its “Ship OS” and “warp speed” industrial tools as part of a broader reindustrialization push in American defense manufacturing. Sankar said usage of Palantir’s Maven defense AI platform is at “all‑time highs,” with the system supporting simultaneous real‑world military events and being pushed out to more combatant commands and edge environments.​

For now, Palantir’s capacity constraints and surging U.S. demand give Karp little incentive to soothe ruffled feathers abroad. He said the company “really doesn’t have the bandwidth to do anything that’s difficult outside of America” and questioned whether European procurement systems are even “load‑bearing” enough to buy “the best product” if it means favoring U.S. vendors over domestic champions.​

At times, Karp sounded almost pitying about his European competition. “To believe you can go and build companies without this is supremely dangerous,” Karp said of orchestrated, production‑grade AI systems. “How do you even perform at half this level is going to be a real question for tech companies and a real question for countries. Can we produce companies that are producing what we produce in a quarter in a year?”

This story was originally featured on Fortune.com

This post was originally published here

When Treasury Secretary Scott Bessent and Federal Reserve Chair Jay Powell convened the chief executives of leading U.S. banks earlier this month to discuss Anthropic’s latest model, Mythos, they signaled a shift in how artificial intelligence is being understood in finance. This was not a meeting about innovation but a warning: that models capable of identifying and exploiting vulnerabilities could pose a material risk to core financial infrastructure.

That concern is justified. But the focus remains too narrow.

In recent years, in discussions with leading financial institutions, I have seen how quickly concern rises once the adversarial uses of AI are understood. Yet the translation into action remains slow and uneven. Much of the current attention is focused on cyber risk. This is a serious threat. But it is not the only one and not the most immediate.

Alongside the risks highlighted by Mythos, a parallel threat is already unfolding at scale. It does not depend on new frontier models, but on AI capabilities that are already widely available. And unlike cyber attacks, which require access to systems, this threat operates by targeting people.

What Has Changed Is Not Just Sophistication — It’s Economics

Artificial intelligence has made fraud dramatically cheaper, easier to execute, and far more scalable. What once required time and coordination can now be automated and deployed at industrial scale. AI systems can generate thousands of convincing messages, voices and videos in seconds, each tailored to a specific individual. This is not incremental. It is structural.

Fraud has shifted from a manual activity to a machine-driven one. Hyper-personalised social engineering campaigns, often powered by AI agents, now operate across multiple channels, jurisdictions, and identities. They impersonate executives, advisers, or family members with increasing credibility, creating urgency and inducing authorised transfers.

In these scenarios, the system is not breached. It is bypassed.

The System Isn’t Hacked. The Customer Is Convinced.

Customers are not necessarily hacked. They are convinced. And because transactions are authorised, existing safeguards are often ineffective. Biometric checks can be defeated by deepfakes. Rule-based monitoring is calibrated to detect human fraudsters, not coordinated networks of AI agents operating at machine speed.

This creates a fundamentally different type of risk.

Unlike cyber attacks, which tend to be episodic and visible, AI-enabled fraud operates as a continuous and distributed leakage of funds across millions of transactions. It is a creeping threat: easier to execute, faster to scale, and often invisible until losses become material. The trajectory points toward trillions of dollars in losses in the coming years. 

The Risk Is Not Only Financial

If the public comes to believe that financial institutions cannot protect customers from manipulation and fraud, trust in the system will erode. The consequences will extend beyond losses. Friction will rise, customers will hesitate, and confidence in banks’ ability to safeguard money may weaken in ways no less damaging than cyber threats.

This is not a greater threat than cyber risk. It is a parallel one. And it deserves similar attention.

A Defense Redesign, Not an Incremental Fix

Most institutions still rely on fragmented data, legacy monitoring and human-led analysis that cannot keep pace with adaptive, AI-driven threats. A meaningful response requires architectural redesign: real-time, AI-native detection; integration of fraud, AML and behavioural signals; and the ability to intervene at the point of transaction, including in authorised payments.

It also requires moving from isolated to coordinated defence. Fraud campaigns target customers across institutions simultaneously, while controls remain siloed. Effective response depends on identifying patterns and campaigns in real time. Privacy and competition considerations remain important, but they can no longer justify structural blind spots. Privacy-preserving technologies offer a path forward, enabling institutions to share signals without exposing sensitive data.

In parallel, institutions need to adopt a “Defence AI” approach: using AI to defend against AI-driven threats. Human-only first lines of defence cannot scale. AI-native systems must support faster detection and response under human oversight.

Regulators Must Convene on This Too — Before the Catastrophe Arrives

The lesson from the Mythos moment is not only that AI can break systems. It is that the financial system is already being exploited in another way: that is less visible, more scalable and potentially just as corrosive.

If the financial system does not respond quickly, the consequences will be severe: rising losses, rising friction, and a significant erosion of public trust.

Regulators should be convening senior financial leaders on this issue, too, as a parallel AI risk, before a catastrophe that is already within reach of bad actors fully materialises. The financial system, the technology sector and policymakers must now recognise the scale of this vulnerability and act with far greater urgency.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

This story was originally featured on Fortune.com

This post was originally published here