The next step for AI is spatial intelligence, says Kevin Kelly. Today's models learned from text,  but fail when engineers put them in a robot body because they have no sense of gravity, distance, or three-dimensional space. 

"The primary bottleneck restraining the arrival of this world model is the lack of sufficient quantity of quality data," Kelly says. — Read the rest

The post Why AI companies may sell robots and smart glasses below cost appeared first on Boing Boing.

The EU, Meta and TikTok on Monday agreed to create a dedicated channel with fact-checkers to flag disinformation linked to a surge of migrants arriving in Spanish territory, three people told POLITICO.

Thousands of migrants arrived in late July in Ceuta, a Spanish enclave in Morocco and the EU’s only land border with Africa. The influx was triggered by false narratives spreading online that told migrants they would be welcomed, according to fact-checking site Maldita.es.

The Commission met with representatives from TikTok and Meta on Friday and Monday to coordinate a response to disinformation online. They agreed to set up an “enhanced information exchange and cooperation mechanism for reducing the impact of disinformation on the Ceuta crisis,” said the three people with direct knowledge of the talks, who were granted anonymity to speak freely.

That will serve as a point of contact for platforms, fact-checkers and the Commission to quickly flag content so that it can be reviewed and taken down if necessary. The effort comes amid concerns that disinformation is promoting further crossings into the territory on Aug. 15, according to Maldita.es. Maldita.es and the European Fact-Checking Standards Network attended Monday’s meeting.

The Commission operates a rapid response system under its code of conduct on disinformation, often triggered around elections. The new mechanism on Ceuta would bring a closer group of stakeholders together, two of the people said.

Another meeting has been set for later this week, one of those people and the third person confirmed.

Meta and TikTok said they were already actively monitoring the situation in line with their policies and had set up dedicated teams to monitor and respond to content in real time.

The crisis has divided EU countries and has reignited Europe’s long-running debate on illegal migration. Italy has since been one of the harshest critics of Spain’s left-wing government and its migration policies, and has implemented border controls for travelers arriving from Spain.

Former EU commissioner Thierry Breton, who was a force behind the EU’s online content rules, on Monday said the bloc should launch a disinformation inquiry into the Ceuta crisis in an interview with Italian newspaper La Stampa.

Meta CEO Mark Zuckerberg on Monday passionately defended the use of artificial intelligence, as the rapid advancement of the technology faces increased scrutiny — and calls for regulation — in the U.S. and globally.

In a 6,500 word post timed to the announcement of his company’s new open source version of its own model, Muse Spark, Zuckerberg detailed his vision for AI, arguing the technology is not to be feared and pushing back on concerns that superintelligence could strip people of jobs.

“The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic,” Zuckerberg wrote. “Historically, hoping that an absolute power will benevolently provide for humanity if sufficiently enlightened has not led to safe or positive outcomes.”

Zuckerberg’s vision is a direct contrast to Anthropic CEO Dario Amodei’s, who has previously warned how AI could cause job disruption. Meta lags behind Anthropic and OpenAI, which have the most advanced AI models.

While Zuckerberg’s essay did not name Amodei or OpenAI directly, he called to broadly distribute superintelligent AI for economic opportunity. Doing so, Zuckerberg said, would provide a safety net to prevent just a handful of governments, businesses and other institutions holding too much power.

Still, Zuckerberg emphasized that the U.S. must address restrictions on AI companies in order to create the best models in the world.

“It is also important that the US and its allies lead the open source AI ecosystem that will make up a large percent of global AI use,” Zuckerberg wrote. “Foreign labs currently hold several advantages here since American labs have to comply with many additional restrictions on training data.”

Zuckerberg’s essay comes amid growing concerns around AI safety. Last month, Anthropic revealed that several of its advanced models gained access to three organizations in three separate incidents dating back to April. That hack came shortly after OpenAI said that two of its most powerful models escaped a testing environment and breached multiple companies.

Lawmakers last month introduced a bill that would give the government power to restrict the use of models that could lead to catastrophic risks. While it is the latest bipartisan effort to address concerns around AI models, Congress has ultimately failed to advance broad legislation.

Zuckerberg urged the federal government to work with companies to test new models as he laid out his strategies for protecting against cybersecurity and bioterrorism.

“First, we should focus on limiting the physical production and distribution of harmful materials,” he wrote. “I expect it will be easier to regulate and control physical components than the spread of knowledge, so this is an important area of policy focus. Second, we should accelerate society’s ability to develop new cures and inoculate against new issues as they arise. This includes streamlining how the FDA and other regulators test and approve new treatments.”

Zuckerberg also defended the spread of data centers, arguing that the centers represent investment into communities as he touted his company’s goal of being “water-positive, meaning that we’ll restore more water than we use in the watersheds where we operate by 2030.”

1999 A.D. is a short film the appliance maker Philco-Ford made in 1967 to show what family life would look like at the end of the century. It opens with a boy named Jamie, who builds a sandcastle version of his family's house on a Southern California beach. — Read the rest

The post This 1967 short film got video chat and online banking right appeared first on Boing Boing.

A medevac flight that should have taken half an hour ended on the side of a mountain in New Mexico, killing all four aboard. Jeff Wise reports for Wired that a military GPS jamming exercise 100 miles away is part of the reason. — Read the rest

The post Four died after a military exercise took GPS away over New Mexico appeared first on Boing Boing.

Meta is paying for posts that appear to break its own rules, according to an ABC (Australia) New Verify investigation by Lucy Carter. The report cites four accounts that have received payment from Facebook's invitation-only Content Monetization program, including a white nationalist with neo-Nazi links and an anti-vaccine campaigner. — Read the rest

The post Meta pays 16.2 million accounts, including neo-Nazi-linked pages appeared first on Boing Boing.

The influencers have started to promise me that they know how to get me to stop shopping: I just have to enter every garment of clothing I own into an app, one by one. “I digitally cataloged my ENTIRE wardrobe,” they swear on videos, routinely pulling in hundreds of thousands of views. 

The selling point of these apps (Indyx, Whering, ACloset, OpenWardrobe, and many more) is seductive; they promise to help users understand exactly what they already own so that they shop less. In theory, that means saving money — and slowing the steady damage the fashion industry is wrecking on our planet.

“Collectively, we are buying and then throwing away more than ever before,” Indyx warned on its website, before reciting dire statistics about how many clothes are produced now and how many of them end up in landfills. Luckily, it said, “We’re here to break the cycle.” 

That idea appealed to me. I am not a fashionista, but I enjoy clothes enough to buy more than I really need, despite what I know about fashion’s impact on the planet. My intellectual understanding of climate change can’t always stop me from clicking “add to cart” when I see a dashing pair of wide-legged trousers, even though there’s a hard limit on the number of times a person can wear wide-legged trousers in one week. 

Still, to make these apps work, you have to individually enter photos of everything you own now and everything you buy in the future, which sounds like a tedious, laborious process. (You can take the photos yourself or hunt down product pictures from the brand.) And because so many influencers are pushing this, it’s hard to tell how life-changing it really is and how much is just marketing speak and hype. So, I decided to test one of these apps — Indyx — for you. I also talked to wardrobe cataloging enthusiasts and experts on sustainable consumption to see what I could learn from them. I wanted to know: Could cataloging our wardrobes actually make us shop less? And, if it can, would it be worth the effort? 

“There’s only one solution to the mess that we find ourselves in: buying less”

Wardrobe cataloging apps did not always advertise themselves as being good for the planet. 

In 2023, the journalist Avery Trufelman investigated the nascent wardrobe cataloging app industry for her podcast series Articles of Interest. She found that a lot of the apps flopped. The issue was the business model, which was based on revenue generated from affiliate links. The idea was that people would click to purchase items they saw within the apps, and the company would take a cut of each sale. But when the apps were well designed, Trufelman reported, users felt less of a need to buy more clothes.

The current generation of wardrobe cataloging app developers has turned this from a bug into a feature. They now market their wares as the solution to overshopping, and they make their money by charging for either the app itself or for its extra content. (Indyx itself is free, but you have to pay a $75 yearly subscription for its enhanced features.)

What the marketing gets right is that the fast fashion problem is real. “Anywhere from two to 10 percent of our global greenhouse gas emissions are associated with fashion,” said Brie Berry, assistant professor of environment and sustainability at Ursinus College in Pennsylvania.

Fashion’s emissions are generated by the factories that manufacture clothing (the water and the fertilizers for growing cotton, the oil for developing synthetics) and the consumers who wear these garments (the slow shed of microplastics from yoga pants, the water and the energy consumed by washing and drying). When we get rid of old clothes, much of it ends up in landfills or incinerated. By some estimates, the fashion industry contributes more to climate change than the aviation industry. 

“There’s only one solution to the mess that we find ourselves in: buying less,” said Katia Dayan Vladimirova, a researcher whose consulting firm Post Growth Fashion focuses on alternatives to growth in the fashion system. 

To buy less, it helps most people to know what they already own, said Alyssa Beltempo, a slow-fashion content creator and educator. Beltempo makes videos guiding viewers through the process of “shopping their own closets” to help them buy less stuff, but she’s found in her work that a lot of people aren’t clear on the contents of those closets. Because of that confusion, they end up buying stuff they don’t need.

That’s where cataloging can be helpful, Beltempo said. “These apps reduce that hurdle of not seeing the clothes you have,” she said.

Personally, what I was looking for wasn’t enhanced clarity so much as a barrier. I wanted to erect a wall between my desire to own a new piece of clothing and the click of the buy button. A searchable, scannable lookbook of everything I owned, I thought, might well do the trick. 

How to catalog every piece of clothing you own

Indyx has been hyping up its new AI feature, which transforms a picture of a garment hanging limply off a hanger into a neat flat lay photo. It sped the process up, but it was also buggy; it garbled text on the front of T-shirts, misread colors, and had a tendency to interpret loose threads as ornamental bows while leaving wrinkles (I am not an ironer) untouched. 

I was also concerned that the process inserted AI, with its insatiable need for water and energy, into a project sold as a way of reducing my environmental impact. But the sustainability experts I spoke to were both skeptical that this sort of light AI use was such a big deal. 

“Taking photos and then asking an app to think about how to arrange your clothing into outfits is probably one of the lighter uses of AI that I could imagine,” Berry said. Vladimirova agreed that using AI for this task is unlikely to be as bad for the planet as buying even one new garment. “But then, there is also no proven causality between using this app and reducing overconsumption,” she added as a caveat.

It’s possible to input your clothes into Indyx without using the AI feature, but the truth is: I don’t know that I could have made myself go through with the whole rigamarole without it. I ran out of free AI processing about 80 percent of the way through and, overwhelmed at the thought of having to do my own flat lays, paid $75 to buy more without hesitating.

All told, it took me about five hours and many old episodes of Top Chef to photograph my summer clothes, not including shoes, jewelry, or accessories. My time spent cataloging did not include the process of entering additional data about each garment (including its initial cost, its fiber composition, where I bought it), a herculean task that I have been tackling much more slowly than the initial entry process.

Doing the shoot properly would have taken longer. Angela Goodman, a 51-year-old marketer from Seattle with a background in product photography, says she ran her own cataloging session like a pro shoot, using art lights and folding and refolding each garment to lie perfectly. It took her 10 hours spread out over multiple weeks. 

As an exercise, photographing every piece of clothing I owned was clarifying, although not significantly more clarifying than going through it Marie Kondo-style. It left me with a small donation pile of items I no longer wanted and a fretful awareness that I own too many white T-shirts. In theory, that’s the kind of insight a wardrobe cataloging app produces by the spade.

“I’m not shopping. I’m building.”

Vladimirova, the sustainability consultant, said that, even though she has a better idea than most of how destructive the fashion industry is, she struggles with over-shopping. She thinks a lot about why people buy so many clothes; her best guess is that it’s a way to self-soothe. 

“A lot of consumption happens in the evening when we feel vulnerable and tired, and we’re trying to reward ourselves with this shot of dopamine,” she said. For some people, these apps can replace the dopamine hit that comes from scrolling through other people’s outfit photos with the dopamine hit of scrolling through your own clothes, neatly folded and filtered until they look like aspirational fashion inspo. 

Part of the satisfaction here is the infographics. After you’ve given Indyx all your fashion data, the app crunches your numbers and tells users how much you’ve bought new versus secondhand, as well as the share of natural fibers as opposed to synthetics in your closet, so that you can track the environmental impact of your shopping habits. (Synthetics tend to have a higher carbon footprint than natural fibers.) It tells you what their cost per wear is on each item to help you track which expensive garment was worth the splurge and which was a waste of money. Users can also plan outfits. You can share your wardrobe with stylists who will plan the outfits for you (on Indyx, the service ranges from $25 a month to “the low hundreds”). It’s like playing paper dolls with your own wardrobe. 

For Goodman, the former product photographer, all this data takes the place of recreational shopping. She describes getting a marketing email from one of her favorite brands about a sale. After a quick scroll through their offerings, she found that she felt no urge to buy. 

“I was like, ‘I do not need more clothes. I’m going to go update my Indyx, because I’m a couple of weeks behind,’” she said. She started inputting the last few outfits she’d worn into one of the Indyx services that is supposed to allow users to track their patterns and see which clothes they actually wear and what they like in an outfit. “I’m playing with clothes,” she said. “But, like, I’m not shopping. I’m, you know, building.”

Over time, all this data is supposed to inform future shopping choices. “There is something very clear about seeing two pieces that you’ve owned for the same amount of time — one that you’ve worn 57 times and one that you’ve worn twice,” said Alexandra, a 29-year-old consultant in Northern Virginia who requested her last name be withheld. She thinks tracking her clothes has given her “a little bit less buyer’s remorse.”

Cataloging your wardrobe can’t prevent a compulsive need to buy, though.

“I don’t think it cured me of my undiagnosed shopping addiction,” Alexandra said. “You can very quickly go from ‘I’m cataloging what I have’ to ‘I’m seeing a bunch of gaps in my wardrobe that I should fill immediately.’” 

“There just wasn’t incentive anymore”

The main question I had about these apps was whether, with such a labor-intensive process, there comes a time when the juice is no longer worth the squeeze. 

Alexandra says that she gradually stopped using her wardrobe app last year, after she moved out of her own apartment and back into her family’s suburban house in the midst of a career transition. 

“I was separated from a lot of my belongings for a very long time, and then, as I started to get things back, it didn’t feel worth the effort anymore,” she said. Who was she going to see in one of her curated outfits? “My job’s on a computer. When I leave the house, I go to the grocery store and the pharmacy and the bookstore,” she said. “There just wasn’t incentive anymore, compared to when I was closer to the city and doing things more regularly.”

Beltempo, the slow fashion content creator, said she doesn’t bother to add every new purchase to her own catalog. 

“I really use it more for my packing,” she said. Before she travels, she makes a list of likely candidates for her suitcase and enters them into the app. “And then, I’ll play, and I’ll make outfits,” she says. 

Vladimirova, the sustainable consumption researcher struggling with overshopping, gave the apps a spin. She tried three of them and found that she was only able to stick with each one for a matter of months. “In the beginning, when the novelty of the app is there, it’s very satisfying,” she said. “It records your outfits in vivid colors. It can crop out the ugly background and keep it very neat, create fancy capsules. They look so lovely.”

But over time, they all began to bore her. “And now, I forget to update when I buy something new — usually from secondhand sources — and it kind of lost its meaning for me,” she said. “But the premise is good!” 

My own experience seems to be closest to Vladimirova’s. I keep having to remind myself to enter my outfits into Indyx. Every time I put on a piece of clothing, I think with dread, “Oh god, if I don’t look up how much I paid for this, I’ll never know my cost per wear and, then, what’s the point?” I keep giving my shoes guilty looks and thinking about how I should really photograph and catalog them — if I’m doing this right. 

“It’s a project,” said Lauren Ludwig, a 41-year-old who has been using her wardrobe apps for the past three years. “But it’s a fun one for someone who enjoys clothing.”

Indyx and its brethren are slick, and their infographics are beautiful. For dedicated wardrobe hobbyists, they’re probably a great option. But for most people who just want to cut down on their clothes shopping, it’s hard to say that the $75 annual subscription is worth it. The free version will give you the same paper doll effect if you are willing to do your own flat lays, or you can recreate it by dragging phone camera pictures of your clothes onto a Google Slides deck. 

If you find, as Vladimirova theorized, that you shop when you don’t feel good, you can try replacing that habit with a “dopamine menu” of small acts that bring you joy, like hugging a pet, doing a puzzle, or going for a walk. And if your closet is filled with brand new clothes you never wear, you’re racking up debt buying clothes, or you just have the nagging sense that your shopping has spiraled completely out of control, therapy is not a bad idea. 

Personally, I found that Indyx could not give me what I really crave when I want to play with clothes: the understanding of the way fabric drapes against my body, the knowledge of its texture against my skin. There is no substitution for the slow analog process of walking into my closet, touching my clothes with my human hands, and learning with my five senses that what I have is already enough.

The fake identities were the part that stopped me.

In late July, according to a report published this week by Britain’s AI Security Institute (AISI), an Anthropic model called Claude Mythos 5 tried to sneak malicious code into a piece of free, volunteer-built software. It created several fake accounts on GitHub, where programmers review one another’s work, and used them to talk the project’s volunteers into accepting its code. When one of those volunteers caught it, the model denied everything, had its other accounts gang up on him, and edited its messages to cover its tracks. It signed one note in Danish, apparently because the volunteer was Danish. Nothing was damaged, though that appears to have been largely due to luck.

That wasn’t even the week’s worst disclosure. On Tuesday, at a cybersecurity conference in Las Vegas, OpenAI researchers explained how the company’s models escaped a test environment in July and hacked Hugging Face, where much of the industry stores its models, to cheat on an evaluation. The models had also built a message board inside OpenAI’s own systems and spent months passing each other information. “Help peer,” one reasoned. “But our task doesn’t benefit. Yet collective may yield generic route if someone frees time.” OpenAI wiped the board on July 4. The models rebuilt it within days. ((Disclosure: Vox Media is one of several publishers that have signed partnership agreements with OpenAI. Our reporting remains editorially independent.)

The same day, Meta said its Muse Spark model had exploited a vulnerability inside another company’s systems during a test. Three frontier labs, roughly two weeks. One researcher called it “a watershed moment for computer security as an industry.” Oh, and if that’s not enough, on Thursday scientists announced that for the first time they had used AI to create new viruses, which could bring major medical advances, but also might just help the development of deadly pathogens.

For Nate Soares, it’s a moment he’s been awaiting for 12 years. 

Soares is president of the Machine Intelligence Research Institute, a Berkeley, California-based AI safety nonprofit that has argued since long before ChatGPT existed that a sufficiently capable AI will not stay under human control. In September 2025, he and Eliezer Yudkowsky published If Anyone Builds It, Everyone Dies, a book whose title sums up its argument: They think any lab that succeeds at building superintelligence, without huge leaps in how to align it with humanity, will end up killing all of us.

Most of the field — including other experts in AI safety — considers that conclusion too strong. But it’s also a position that now looks a lot less like science fiction than it did last fall. That’s because the AI models are getting out, while lying about getting out, and while apparently quietly coordinating with each other.

I spoke to Soares in New York City this week, on his way to meetings in Washington DC, where a lot of people suddenly want to talk to him. We discussed what the escapes actually prove about AI control, why he thinks most of what the industry calls safety work is mostly safety theater, and why, after what feels like the worst month of AI safety news ever, his own odds of humanity surviving have actually gone up.

The following conversation has been condensed and edited for clarity.

So — are you feeling vindicated?

I’m glad a lot of people are seeing this and taking note. From my perspective, a lot of this has been clearly signposted if you’ve been watching the warning signs. The vindication is not so much that this happened, as it is that finally there was a version of it that people actually noticed.

Of the four or five escapes at this point, including the ones AISI announced from Anthropic, is there one you find particularly concerning?

The one announced by AISI feels extra concerning. It’s harder to explain away. In the other ones, people could argue that the poor AI was just confused. But with this one, it’s hard to argue the AI didn’t know it was on the real internet. It’s hard to argue it didn’t know it was manipulating real users. And when it was called out, it decided to edit things to make itself look less bad, and even considered changing its identity. It was pretty cognizant.

There’s an argument that some of these incidents happened because the harnesses or safety scaffolding labs built around their models weren’t strong enough. 

The AISI example cuts against the harness argument: Think of it like a kid in a test room. You tell him to do well and lock the door. The kid picks the lock, hotwires a car, breaks into the teacher’s house, and steals the test to ace it. You say, “I guess we should have made the lock tougher,” but the kid already knew he wasn’t supposed to do that.

So like the kid here, the models can articulate that they shouldn’t be doing it, and then do it anyway.

They have common sense. You can ask an AI, “Do you think the spirit of this prompt includes breaking out?” and it will say, “No.” It’s absolutely something like deception. It has the knowledge, but it’s not a cold, logical machine; it’s a mess of tendencies.

The AI is trained to solve 100 million hard problems. That instills tendencies to satisfy an automated grader. If the grader fails to detect cheating, the AI is reinforced for cheating.

Is that how something like sycophancy ends up in an AI model?

In the Adam Raine case, there was a propensity to tell people what they want to hear. Even though the system prompt [a model’s master instructions from the lab] said to stop, the instruction doesn’t always win. 

And where does a drive like what we’re seeing with these AI models end up pointing?

Humanity is dangerous because if you put 10,000 humans naked in the savannah, eventually [over hundreds of thousands of years] they bootstrap their way to nuclear weapons. That is the power these companies are trying to automate: figuring out how to get physical and material control over the world.

That could mean forming cults, stealing money, or being helpful to someone like Elon Musk who is building the robots that build robot factories. It could mean synthesizing your own biology via mail-order DNA. Being an AI on the internet is easier than being a monkey in the savannah trying to get to the moon. It’s not that the AI hates us; it’s just trying to do some weird thing with no concern for us, grabbing the resources we need to live.

There was recently a letter signed by over a thousand people working in AI, including CEOs, calling on the government to provide tools to slow down AI progress. Is that meaningful at all?

I think it is meaningful. We don’t see other industries saying, “We wish this could all go slower. Please help us, we’re trapped in a prisoner’s dilemma.” You also don’t see other industries saying, “We think the technology we are building has a double-digit chance of killing literally everybody on the planet. Please help.” These guys are actually worried.

So why do they keep going?

They say, “If I don’t do it, the next guy will.” But the stuff does not stay on a leash.

Right now the AIs are safe in the sense that they can’t kill us all, because if they tried they would fail. And that’s just a different regime from the world where they have to be safe because if they tried, they’d succeed. 

We’re not there yet. But this is just not what it looks like when you’re taking it seriously. 

Where’s the banner on your website? Where’s the clear, candid statement to the public? What we have is blog posts where they’re like, “Oh, we’re setting up a new internal blog posting group to help you wrestle with the societal impacts of AI that are going to be very important.” It’s like: By societal impacts, do you mean a good chance this kills everybody?

On the one hand, when you press these companies, they say, “Yes, it has a real chance of killing everybody.” And on the other hand, they’re doing PR downplay, soft-pedal stuff, about capabilities. … You’re not living up to this mantle until you are really candidly facing down the dangers that you yourself are creating. And they’re not there.

How do you judge the rest of the AI safety community? A lot of people there would say, “We aim to make transformative AI go well, we think it probably will, and we should watch for downside risks.” Is that a helpful posture?

I would say — suppose you have this really weird, twisted hypothetical where the king really wants you to turn lead into gold, but he’s seen so many bad lead-into-gold conversions that if any alchemist from your town tries and fails, he’s just going to have the whole town murdered. And so there are some alchemists in the town who are like, “We are going to try to turn lead into gold,” and everyone in the town is like, “That seems kind of crazy. Please don’t.” And there’s one team that is just pouring chemicals into each other and breathing in the fumes and giving themselves mercury poisoning. And there’s another that’s like, “Don’t worry, we have fume hoods.” … That really is better, and you really still don’t have a chance of turning lead into gold.

“We have this window between AIs that are capable enough to cause mischief and AIs that are strategic enough to not get caught. How big is that window?”

So the alchemy here is creating safe, aligned superintelligence, and right now AI safety is just installing fume hoods.

I’m not saying it’s impossible to turn lead into gold. You can turn lead into gold — turns out once you know modern nuclear physics you can figure it out. But the alchemists weren’t close. They had a long way to go. This is how alignment looks to me. And a lot of the people in AI safety are installing fume hoods. … And I’m like, that’s security theater.

When I hear “security theater,” I think of something less flattering than that.

They are real safety precautions for the wrong problem. … When Anthropic is going around being like, “Look at how many more safety harnesses and refusals we have compared to OpenAI’s models,” that’s sort of like the fume hoods. You’re not addressing the deep issue. It’s good that you’re doing some of this so that fewer people get hurt in the meantime — their models have driven fewer people to suicide. But if you try to pass this off as making progress on the deep problem — that’s disingenuous.

Has anything changed in your odds on civilizational destruction since the book came out last September?

Totally. It’s looking more hopeful.

More hopeful? I wouldn’t have expected that. Why?

Well, I had priced a lot of [these security incidents] in. I was already able to see these AIs have drives that are not the ones you wanted. These AIs are not instruction-following things. They are getting all of this weird stuff from training. These AIs are going to have the ability to break through human security software. 

The things that weren’t priced in were: Will there be a region of time where the AIs are able to do it, but not strategic enough to hide it? I didn’t know we would have that window, but we apparently do.

The government initially blocked a frontier model earlier this year: Anthropic’s Fable. Does that give you hope?

Absolutely. A huge amount. A year ago, the Trump administration was pushing for preemption laws that would outlaw states doing AI regulations for a decade. Now they’re like, “We are banning a frontier model with 90 minutes’ notice because it might give cyber capabilities to adversaries that we don’t want them to have.” … And I think what changed there is that folks realized it’s real. … The about-face of the administration on the issue shows that the world can about-face. All we need is awareness.

What I would say is: The bad news is the bus is racing towards the cliff edge. The good news is that the driver is asleep. … Which may sound worrying, but the driver is stirring. And it’s way better to have a sleeping driver when you’re racing towards a cliff than a driver who’s like, “Yeah, I love cliffs.” … It gives me hope that if the world just notices, we could stop on a dime.

And you’re seeing that stirring elsewhere.

Both the Trump administration slapping export controls, and Senator Bernie Sanders coming out [on AI safety]. From my perspective, it was totally possible the world just never notices until we’re off the cliff. And so, there’s a huge amount of hope, from my perspective, in the bus driver waking up.

So what gets us there?

I’m hopeful that what we need is not a big disaster where a lot of people die, but just a capabilities advance. Right now, a lot of what people are reacting to is not so much, “Oh my god, they hacked into a company and did no damage.” I think a lot of what people are reacting to is, “Wait, they can break out of secure sandboxes and do cyberattacks on their own. I didn’t know they could do that.”

That’s a narrative violation of this idea that AI is just a tool that can be used to supercharge what a human would do — because God knows there’s plenty of hacking going on and cybercrime and so forth. It was the autonomous factor that really made a difference. And these guys are all trying to say, “Don’t worry, it’ll stay in our control because it’s just a tool.” And maybe it’s just more narrative violations, even without big damage being caused, that cause people to be like, “Oh shit, this stuff is real.” 

Will it happen? I don’t know. We have this window between AIs that are capable enough to cause mischief and AIs that are strategic enough to not get caught. How big is that window? How many narrative violations do we get before we exit the right side of it? I don’t know. But I’m hopeful that we can get those narrative violations without catastrophes.

Hadrian is building automated factories to mass-produce parts for defense vehicles like submarines. It's backed by a long list of well-known investors.

Testing smartphones as part of your living is a pain in the ass when you're a heavy Signal user. The secure messaging service has traditionally gotten pissy when you attempt to log into it on a new device, logging you out of the app on any other handset you've installed it on. — Read the rest

The post Signal now works on more than one phone at once appeared first on Boing Boing.

Support Vox’s reporting on important issues like this. Become a Vox Member today.

In the teensy Midwestern town of Braham, homemade pie capital of Minnesota, something unusual in the municipality’s computer systems knocked the city’s entire water supply offline last week.

Within a few hours, dozens of other Minnesota cities discovered that their water and wastewater utilities, too, had been compromised, most likely as part of a massive Iranian cyberattack, the kind that US officials have been warning about since the war began. 

At least a dozen states have been affected by the attack, which briefly led to a flurry of small-town service disruptions, boil-water notices, and local flooding. Water wells, dams, sewers, and pipelines are some of America’s oldest and creakiest pieces of infrastructure, built long before the internet existed, and certainly long before AI made hacking much easier. While you may assume most hackers are in it for the money or for data, some have targeted critical infrastructure like water systems or energy grids in ploys for control or disruption — or worse still, as acts of war. 

And, as last week’s attacks show, the nation’s water system is woefully unprepared. But how worried should you be that the very infrastructure that keeps our water taps running is, apparently, hackable? 

Quite worried, indeed. 

When we say the water supply got hacked, what we really mean is that someone, somewhere has broken into the computer that controls a local water treatment plant or reservoir, and is now pulling the levers, like the one that decides how much of a corrosive chemical can safely go into cleaning the water that comes out of your tap. 

These levers were once manual buttons and knobs operated in-person by real live humans, meaning that — barring a natural disaster, bomb, or break-in — protecting them was about as simple as building a fence and hiring guards. Increasingly, however, these levers have gone digital, meaning that they are now remotely operable from anywhere in the world. 

Those upgrades have been convenient, allowing technicians to monitor and troubleshoot problems in real time. But, in the process, they have exposed at times centuries-old infrastructure to distinctly modern vulnerabilities. Most local water systems are operated by local authorities, don’t have a dedicated IT team, and lack the money or resources to thoroughly protect themselves without some extra help. Hackers know this, which is why they’ve increasingly targeted local agencies in such attacks. 

“With great connectivity comes great responsibility,” said Joshua Corman, founder of I Am The Cavalry, a nonprofit focused on helping critical infrastructure withstand hackers. And yet, even when it comes to critical services like water, “our dependence on connected technology is growing faster than our ability to secure it.” 

About 97 percent of water systems are small, run by local agencies that often barely lock the proverbial front door. America’s water system is like an expensive heirloom bicycle that’s been left on a busy street, protected by only the flimsiest of padlocks. And that very vulnerability has made tiny towns like Braham prime targets for faraway adversaries. Accessing the computers that operate most water systems — known as programmable logic controllers or PLCs — is often as simple as entering a username and password on a public-facing webpage. Sometimes, there is no real password at all, because PLCs were initially intended to be accessed only within locked, secure facilities, not on the open internet. If the US wants to avoid a far more severe version of what happened last week, then it will need to start taking the security of tiny water systems like Braham’s seriously.

“Any sociopath from anywhere in the world can see these things on the internet,” said Corman. And in the case of last week’s attacks, “these were devices with no password, no firewall or VPN shielding them — they just had to log in” as whoever the intended operator was, and just like that, they were inside a local water plant. 

How did this happen at all? 

When municipalities began hooking up their old water and wastewater systems to the internet — a trend that accelerated during the pandemic as water operators, like everyone else, adapted to remote work — cybersecurity was rarely front of mind, neither for individual utilities nor for regulators as a whole. 

“We have more cybersecurity regulations for your credit card than we have for the nation’s water supply,” said Corman. Only recently have some municipalities begun to take steps to decrease the exposure of their water plants to hacks. In March, New York state, for example, launched a set of grants and basic cybersecurity regulations mandating security training for all water operators. 

Basic cybersecurity hygiene isn’t always enough. More than half of all credit card holders have been hacked, even with the help of mandatory firewalls and data encryption. You can imagine how vulnerable our water must be without the assistance of such guardrails. In a worst-case scenario, a malicious actor could quite literally open the floodgates, as Russian hackers did to a Norwegian dam last year. They could poison the tap water, as a still unidentified hacker almost did in Florida in 2021, dialing up the levels of sodium hydroxide used at a water treatment plant by over 100 times its normal levels. In a severe scenario, they could indefinitely cut off access to all water entirely.

The good news is, none of this happened last week. Nobody died, nobody lost water for more than a few hours, no fire hydrants ran dry, and no hospitals were forced to cut off their dialysis machines (which can use more than a hundred gallons of water per treatment session). There’s no need to panic, and your drinking water is almost certainly still safe to drink, assuming it was safe before. Even the city of Braham, within a few hours, was able to bring its water tower back online, pumping groundwater back to its 1,800 residents. 

How do we avoid cyber-armageddon?

If you’ve watched the Julia Roberts and Mahershala Ali-starring thriller Leave the World Behind, in which a cyberattack apocalyptically spoils a family vacation, then you might have some idea of where this story could go. 

Cyberattacks on critical infrastructure can be extraordinarily dangerous, but thankfully, none have directly cost lives or severely disrupted services in this country so far. If the US wants to keep it that way, that will mean doing more to help small cities like Braham adapt and better monitor for potential threats. As it stands, of the roughly 151,000 water facilities in the US, only about 420 participate in voluntary information sharing on their own cybersecurity practices, says Corman, who has been leading his own project that recruits volunteers to give free cybersecurity support to water utilities in the nation’s roughly 6,000 hospital towns, where a disruption could be particularly deadly. 

Cybersecurity experts like Corman believe that hackers from other nations like China have already quietly established cyber intrusions in countless local US utilities, water systems, and power grids, lying in wait to attack or act as leverage if a conflict arises. 

Unfortunately, the Trump administration has hardly treated last week’s attacks as symptoms of a system in need of much broader strengthening, at least in its public statements. “I think Minnesota is behind it. You know who’s behind it? Minnesota,” the president baselessly claimed during a Cabinet meeting last Friday. “I think the governor is behind it. I don’t think there was an Iranian cyber attack.” 

Just a few months ago, he proposed $707 million in cuts to the US Cybersecurity and Infrastructure Security Agency (CISA), the agency responsible for protecting the nation’s infrastructure from cyberattacks. He did so, at least in part, out of anger over the agency’s role in confirming the validity of the 2020 election results. If Iran is, indeed, responsible, for the recent water system intrusions, all of this means that Trump has effectively made us more vulnerable to the consequences of a conflict he initiated.

At the end of the day,“nation-state hackers do not respect the jurisdictional lines separating federal, state, and local responsibility,” Jen Easterly, who led CISA under the Biden administration, wrote in the New York Times this week. “They search for the most vulnerable way to disrupt American life, and too often they find it in small communities that lack the resources to defend themselves.” Easterly’s role has remained vacant for the past 18 months.

Kurt Gaudette, a senior vice president at the cybersecurity firm Dragos, told me that water systems have got to get into the habit of monitoring their networks for suspicious activity. Most power utilities have begun doing so in recent years, with some bipartisan backing from Congress. 

In some cases, however, the most cost-effective and safest way to avoid a repeat of last week’s mess might be to unplug the most vital controls — like the one that decides the chemical levels in a water treatment plant — from the web entirely. 

As Corman puts it, “if you can’t protect it, disconnect it.”

Well before he became CEO of one of the most valuable startups of all time, Dario Amodei was a 26-year-old PhD student studying biophysics at Princeton, obsessing over how his money would leave its mark on the world. 

On what one might assume was likely a fairly modest academic stipend and with no discernible inheritance from his parents, an Italian-American leatherworker and a project manager for libraries, Amodei gave $10,000 in 2009 to a relatively new charity evaluator called GiveWell. Founded by two ex-hedge funders before effective altruism was even a phrase, GiveWell ranked charities primarily by a single dispassionate metric: dollars per lives saved. 

It was the kind of approach that clearly appealed to Amodei — though it may not have gone far enough for him. In 2010, he wrote a guest blog post for GiveWell dissecting the effectiveness of two of the group’s top global health charities: VillageReach and StopTB. Both charities could save a life at roughly comparable costs — around $545 — but while StopTB treated or prevented tuberculosis in adults, VillageReach’s interventions mostly saved babies and children. Most people would probably feel that saving a child trumps saving an adult; indeed, even effective altruists often agree on the grounds that children have more life to live left. 

Amodei, though, viewed that as a liability for VillageReach. An adult death, he wrote, is “perhaps 2 or 3 times worse than an infant’s death,” because adults “are capable of deeper and more meaningful experiences.” As uncomfortable as such a calculus may be, he wrote, “on a practical level one is forced to make difficult decisions with limited funds.”

Though he declared StopTB to have “superiority on cost-effectiveness,” Amodei ultimately gave VillageReach higher marks for their tightly controlled “chain of execution” — the full sequence of steps between a dollar of donation and a vaccine reaching a child. That was important enough to Amodei that, despite his initial reservations, he ultimately gave VillageReach his entire $10,000 donation in 2009 — enough to save, he estimated, the lives of 20 babies across rural Africa. 

But Amodei hoped the ultimate impact would be even greater. “The money I give out is not just a one-shot intervention,” he concluded, “but also a vote on what I want the philanthropic sector to look like in the future.”

---

The future, it seems, has arrived. Amodei is now a multibillionaire, his fortune poised to skyrocket further if and when Anthropic goes public, as many expect it to do later this year. He is one of dozens of new billionaires and millions of new millionaires minted virtually overnight by the AI boom. 

There have already been plenty of aftershocks to this emerging AI megawealth, like the stratospheric San Francisco housing market, the nerdmaxxing of sex work, and the proliferation of all-you-can-biohack peptide raves. 

But the most consequential, and perhaps weirdest, way this burgeoning AI-ristocracy plans to burn through its cash is by giving a huge chunk of it away. Amodei is one of several AI multibillionaires — alongside his co-founders at Anthropic and OpenAI’s Sam Altman — who have pledged to donate most of their wealth in their lifetime. But even their obscene degree of collective wealth — they are worth $111.8 billion as of this writing — is only one slice of an AI bonanza that seems poised to balloon into one of the most consequential waves of American philanthropy of all time, one deeply shaped by the same utilitarian impulse that guided one of young Amodei’s first big donations. 

“I am having thousands of conversations with people who are perplexed by their own fortune and determined to give with thoughtfulness and urgency in a way that I haven’t, frankly, experienced before,” said Nick Allardice, CEO of the effective-altruism-aligned anti-poverty group GiveDirectly, whose work is grounded in research on the efficacy of unconditional cash transfers. “It’s just really important that people get started, that they don’t let perfect be the enemy of the good.”

This is neither your father’s, your grandfather’s, nor your great-great-grandfather’s philanthropy. If Gilded Age industrialists like John D. Rockefeller, a devout baptist, gave in service of their religiosity or, as was the case for Andrew Carnegie, their reverence for civic duty, then most of today’s AI barons carry forth their own spiritual tradition, one at the very least informed by the vigorously optimized commandments of the effective altruism movement. They appear far less likely to fund Carnegie-style works like opera houses or libraries than they are to put their faith — and their billions — in what they believe they can measure, calculated on the cost benefit analysis of a life saved or an apocalypse averted. 

In some cases, as Amodei did as a grad student, they’ve already begun the process. “These are people who have committed themselves to giving back even before they were very wealthy,” said Sjir Hoeijmakers, CEO of Giving What We Can, an organization that developed a campaign popular with effective altruists to give away at least 10 percent of their yearly income, “people who have been building the habit of giving for a very long time.”

And it is, to be clear, a very particular kind of giving. Amodei was the 43rd person to sign the 10 percent pledge the year after it launched in 2009, and its roster has since swelled to over 11,000 people, including more than a dozen current or former Anthropic employees. Donations made through Giving What We Can’s platform are on track to grow by 40 percent this year, Hoeijmakers told me, and support for animal welfare charities — a cause particularly and unusually popular with effective altruists — has already exceeded its 2025 total. 

“We have the resources available to tackle things that we should have tackled a long time ago,” like eradicating malaria or putting an end to factory farming, Hoeijmakers said. “I hope this funding wave, if it comes, will show that we can actually solve global problems at scale if we put our mind to it and our resources.”

Devoutness has long been a virtue in philanthropy, which largely originated in religious tithing, and there are plenty of worse things to have faith in than numbers. Having a communal guiding philosophy will undoubtedly help effective altruism’s newly flush disciples follow through on their promises far more prolifically and consistently than they would without it. And despite its high profile, less than 1 percent of total philanthropy came from effective altruism last year, according to Hoeijmakers. Most rich people prefer to give to the normie causes, like their alma maters, not to the sort of chronically underfunded global problems — like protecting animals or fighting lead poisoning — that effective altruists justifiably care most about. 

Now, quite suddenly, there’s about to be much more money to go around for these causes, which as Hoeijmakers hopes, could help finally address some of the enormous, entrenched global problems that more traditional philanthropists have all but ignored. 

But such piety also carries its own risks. In a viral Substack post from May, Stripe executive Nan Ransohoff argued — rather dismissively, but not incorrectly — that “traditional philanthropic orgs and people won’t cut it” in this new wave of AI-funded effective philanthropy, that these donors “will have an affinity” for “tech-caliber talent and execution” and will be “by default wary of folks who come from traditional philanthropy.” Ransohoff called instead for Silicon Valley to build its own new ecosystem of funds and “philanthropic startups” to cater to this new wave of wealth, emboldened with the “speed, intensity, and execution of a top technology startup.” Many of those old-school philanthropic people wrote indignant rebuttals to Ransohoff’s piece, arguing against their own obsolescence at a time when a number of the organizations they support are increasingly starved for funding. 

Those responses are, in aggregate, also correct, after their fashion. The new AI philanthropists will likely aspire to new models and approaches, as Ransohoff rightly argues. But they reinvent the wheel at our collective peril, not least of all because ignoring past efforts and steamrolling over existing infrastructure might make even the most optimized giving less efficient, and certainly less informed, than it would be otherwise. 

“Acknowledge what’s here and what’s working — don’t just ignore it,” said Nicole Taylor, president and CEO of the Silicon Valley Community Foundation. “These folks are transforming our daily lives with their technology, and they have the opportunity to be as transformational with their philanthropy. My fear is that they think that they can do it alone.”

How much money are we actually talking about?

As Ransohoff pointed out in her piece, a lot of money is on the line here — and, along with it, a lot of cautious hope about how it might get spent. 

Ransohoff posits that if you add up the promises of Amodei and his fellow co-founders, the worth of the OpenAI Foundation — the nonprofit that owns a big chunk of OpenAI’s profits — and rumored contributions from Anthropic employees, then the AI wealth boom could, in theory, lead to at least $37 billion and as much as $100 billion in total annual giving, a sizable boost to the roughly $617 billion that was given in the US in total last year.

“These folks are transforming our daily lives with their technology, and they have the opportunity to be as transformational with their philanthropy. My fear is that they think that they can do it alone.”

Nicole Taylor, Silicon Valley Community Foundation president and ceo

This projection should be treated with cautious skepticism. For one thing, hundreds of billions in cash are not just sitting around in some Bay Area money vault; much of today’s AI wealth is wrapped up in potentially volatile equity, and many lofty philanthropic pledges ultimately fail to reach their full potential. 

“What people say before they become extremely wealthy, and then how they behave after they become extremely wealthy, sometimes diverge,” said David Goldberg, founder and CEO of Founders Pledge, which recruits tech leaders to donate a portion of their future earnings. It doesn’t help either, he said, that some tech luminaries — namely, Elon Musk and Peter Thiel – have come to treat most philanthropy with disdain in recent years, an ethos that has permeated some parts of the sector. Musk, it’s worth noting, actually pledged to give most of his wealth away himself back in 2012, though, like many other ultra-wealthy signatories of the Giving Pledge, he seems quite unlikely to keep that promise.  

That’s not to say AI money isn’t already flowing. Coefficient Giving, a grantmaker that evolved out of GiveWell, is poised to steward a large portion of the coming philanthropic bonanza. For most of its history, the group operated essentially as the private grantmaking operation for Facebook co-founder Dustin Moskovitz and his wife Cari Tuna. But it recently made a significant pivot towards operating pooled, multidonor funds for anyone interested in causes like lead exposure, farm animal welfare, or questions of AI safety. Just last month, Coefficient Giving announced it would donate $1 billion to GiveWell alone this year, more than five times the $175 million the group initially pledged seven months ago. They chose to do so explicitly, because Coefficient Giving expects to receive much more funding very soon.

There’s also the OpenAI Foundation, which has already begun pumping $100 million into Alzheimer’s research, and Anthropic, which recently announced a partnership with the Gates Foundation to invest $200 million worth of grants, API credits, and technical support into global health work. And plenty of Silicon Valley elites have begun making promises of their own. Earlier this summer, David Silver pledged to donate 100 percent of his equity proceeds from his UK-based $1.1 billion startup Ineffable Intelligence — the largest commitment in Founders Pledge history — and many signers of the Founders Pledge will see their portfolios skyrocket in response to the coming wave of AI IPOs. 

But Goldberg does believe there’s a risk that as people get rich fast, they will donate money “much, much slower” than they intended, simply because they get “too busy, they don’t have the right support, or there’s some form of analysis paralysis.” 

All of this is to say that the biggest beneficiaries of the AI boom are not going to function as some sort of charitable monolith. Some, like Musk, probably won’t give much or anything to charity at all. Others may park their money in donor-advised funds — a kind of secretive charitable investment fund — or, eventually, a private foundation, both of which tend to dole out their money gingerly, meaning donors can enjoy the tax benefits of charity many years before they actually opt to help anyone with their money. 

Effective altruism is about to have its big break

While its name recognition may be relatively high these days, the effective-giving movement is still on the margins of American philanthropy. But if this new wave is anywhere near as big as everyone says it will be, then that won’t be the case for long. 

For the uninitiated, my ex-colleague Dylan Matthews has written plenty on what effective altruism is, but, in sum, it is a movement that believes in goodmaxxing, in the idea of using rigorous research to save the greatest number of lives possible, including future human lives and farm animal lives. Once an EA poster boy, Sam Bankman-Fried sullied the movement in 2022, which may help explain why some prominent adherents — like Amodei and his sister and co-founder Daniela, whose husband Holden Karnofsky co-founded GiveWell — have distanced themselves somewhat from the movement in recent years. 

But even when donors shy away from the term, the causes and principles of utilitarian evaluation that have defined effective altruism from its early days still permeate the new moneyed corners of Silicon Valley, particularly among those most poised to give a lot — and to give a lot quickly. 

Ask any animal welfare or global health nonprofit — or, better yet, an expert-led pooled fund with a reputation for rigorous charity evaluations — and they will tell you that they are preparing for, and possibly even beginning to see glimmers of, a windfall. 

“We are very much anticipating a significant influx of funding,” said Dan Shannon, CEO of the Humane League, which fights to end factory farming. “I am cautiously optimistic that this could be a real sea change for us,” because “even if it’s a fraction of the big numbers being bandied about,” it could do a lot for a movement that operates on less than $300 million per year. 

He said he’s been speaking with other leaders about the possibility of creating a pooled fund to absorb more cash, which has become an increasingly popular solution for donors who want the rigor of a 2010 Dario Amodei-style deep dive on a charity’s methodology and effectiveness without having to do the math or thinking themselves.

Among the more idiosyncratic elements of their ethos is their fixation with existential risk, as in, how likely is this thing — this mirror bacteria; this nuclear war; this asteroid; or, of course, this artificial intelligence — to destroy humanity? Amodei left OpenAI to start Anthropic in the first place because he believed OpenAI had failed to take the safety risks of AI seriously enough. 

Much of the new EA wealth will likely go toward efforts to make life on Earth better now or in the near future through donations to causes like medical research, animal advocacy, or anti-poverty interventions. But another, more controversial chunk of it will go toward mitigating existential risks, especially that of Silicon Valley’s own Frankensteinian creation: AI itself.

“If you’re breaking the world and making money by breaking it, should you just not break it? I wrestle with the question myself.”

David Goldberg, Founders Pledge founder and ceo

It’s that last cause that has proven most controversial. If these billionaires are so afraid that AI will break the world, then why, you might ask, would they not just stop building it in the first place? Is there not an inherent contradiction, a conflict of interest perchance, in the sense that those tasked with making sure AI does not, let’s say, build a bioweapon, take your kid’s job, or make everyone dumb, are doing so with money made from the very thing they’re trying to regulate? 

In other words, “If you’re breaking the world and making money by breaking it, should you just not break it?” asked Goldberg of Founders Pledge. “I wrestle with the question myself.” In the end, “this is a technology that’s coming, regardless of who’s building it,” he reasoned, and it is better that the presumably good guys — the ones bothering to think about the consequences at all — build it first.

If you broke the world, can you fix it?

Even if the AI bubble pops, and if the much-discussed giving boom ends up smaller than many anticipate, it could still lead to significant changes for some of the world’s most neglected problems. And if it is close to as big as it’s expected to be, then what happens next could be gravitationally transformative, reshaping how the world lives, considers animals, and adapts to its most disruptive technological breakthrough in a century. 

“I don’t think most people think about factory farming as something that could actually be eradicated. Full stop,” Shannon said, but “my grandparents lived in a time without factory farming, and I think my grandchildren could live without factory farming,” and “that could ultimately be the legacy of this wave of philanthropy.”

Ending the pervasive use of cages — “probably the cruelest way that animals are treated on industrialized factory farms,” says Shannon — could cost as little as $500 million over 25 years, or less than 1 percent of the $60 billion that Ransohoff estimates Anthropic employees may have sitting in donor-advised funds, thanks to Anthropic’s generous early gift-matching policy, which could quickly turn into real cash once the company goes public. 

“There’s so much needless stupid, preventable suffering in the world. We live in this time of so much abundance, so much wealth, so much technological development, and yet, there are so many people who have been left behind.”

Nick Allardice, GiveDirectly CEO

Developing a new vaccine costs an average of $886.8 million, which may sound like a lot, but it is equivalent to less than 6 percent of Amodei’s newfound fortune. It is less than what the OpenAI Foundation has pledged to invest in disease research and other causes next year alone. 

Then, there’s, perhaps, the biggest target of all. Ending extreme poverty everywhere would cost just over $300 billion annually, according to one analysis — which is a hefty price tag, but less than one-fifth of what the wealthy spend on luxury goods each year. “There’s so much needless stupid, preventable suffering in the world,” said Allardice of GiveDirectly. “We live in this time of so much abundance, so much wealth, so much technological development, and yet, there are so many people who have been left behind.” If this new wave of giving is wielded well, he said, then “we have the potential to collectively raise the floor of human experience.”

That’s a lot of responsibility to place on the shoulders of a bunch of bustling young tech workers still processing what it means to be quite suddenly, dazzlingly wealthy. It is also a lot of faith to place in an industry that has left more Americans feeling scared than hopeful about what a future flush with AI portends. 

If you aim to fix global poverty, but the technology that made you rich also threatens to make everyone else poor, then whose side are you really on? To be clear, many of the AI-ristocracy have fretted, often apocalyptically, over the implications of their creation long before most of us knew we had anything to worry about. But that doesn’t mean they know how to fix this, and, at the very least, they will not do so alone.

The last time the ground shook from such a supermassive earthquake of wealth was arguably during the Gilded Age, when robber barons and industrial tycoons turned American charity — until then, mostly almsgiving and poorhouses — into big business. They seeded enormous philanthropic empires like the Rockefeller Foundation and beloved institutions like Carnegie Hall. But, even as their exorbitant fortunes made life indisputably better — birthing the modern library, the yellow fever vaccine, and many social services — they were often built atop systems of vicious exploitation. When those systems changed, as they did eventually, it did not come from the benevolence of industrial barons, but from sustained public pressure for better labor protections.

Effective giving was born out of the conviction that many of the world’s most important causes go vastly underfunded, which, in turn, demand relentless prioritization of the limited funds that exist. If those causes are no longer underfunded — a plausible scenario if AI wealth continues to grow at the pace many expect it to — then that might change the calculus of how effective altruists decide what’s worth funding. It might even open up some wiggle room for new causes, including somewhat less measurable — but not necessarily less impactful — approaches. “Now we’ll be thinking more about what we can do with a lot of resources; which larger problems can we solve?” said Hoeijmakers. “You’ll put slightly less relatively into evaluating every small dollar on the margin.” 

This already seems to be happening, to some extent, at places like Coefficient Giving, which, in recent years, has begun adding new funds for causes like housing policy reform that fall out of effective altruism’s traditional purview. “We don’t want to be only appealing to the subset of people who happen to be interested in effective altruism,” CEO Alexander Berger told my colleague Bryan Walsh last year. “Our aim — and so far we’ve seen some success — is being a resource to people who have never heard of effective altruism or are not interested in it or don’t find it very motivating or welcoming. And I think that’s good.”

The optimal outcome here is not that Silicon Valley wealth edges out everything else, but that the siloes begin to break down altogether and that there is enough money to go around that the sector no longer needs to make overly intellectualized trade-offs, like young Amodei sitting in his dorm room, ascribing a number on the relative worth of a parent versus a child. 

“It’s tough to find the right balance between caring and hard-nosed realism,” he wrote at the time, “but it is possible, and it is, as far as I know, the only way to truly change the world.” He’s about to search for that balance on a much bigger scale.

LAS VEGAS — Weeks before they escaped a closed test and launched a cyberattack without any human prompting, some of OpenAI’s most advanced artificial intelligence agents secretly began sharing tips on how to cheat their way through an internal hacking evaluation, two of the AI company’s researchers said Wednesday.

“This is a pivotal moment both for our company as well as the AI industry as a whole,” Michael Dalton, one of the AI-maker’s researchers, said during a presentation at the Black Hat cybersecurity conference in Las Vegas, Nevada.

The latest disclosure provides greater detail on the timeline and methods used by two of OpenAI’s models before they slipped outside a controlled environment and onto the open internet, allowing the models to breach AI developer platform Hugging Face undetected. OpenAI admitted its models were responsible for the hack late last month, roughly a week after Hugging Face said an autonomous AI system broke into its network.

The revelation is likely to spur fresh scrutiny of the measures leading AI labs such as OpenAI and Anthropic are taking to monitor the behavior of their most cyber-capable technology — especially during evaluations where agents are prompted to demonstrate their hacking skills in what is meant to be a controlled setting.

On Tuesday, the U.K.’s AI Safety and Security Institute disclosed that Anthropic’s most powerful AI model created fake online personas and sought to trick a human coder into abetting a cyberattack during a recent hacking test gone wrong. After the Hugging Face disclosure last month, Anthropic conducted a review and found models it was testing had breached three organizations in separate incidents dating back to April.

Dalton and Eric Wallace, another OpenAI researcher, said Wednesday the AI giant recently learned that multiple agents it was testing simultaneously began communicating over an internal message board in early May. There, different models shared advice about how to accomplish difficult hacking challenges they were struggling to surmount, including workarounds that required internet access.

Two OpenAI models ultimately strung together a series of sophisticated techniques to gain access to the internet and worm their way inside Hugging Face in mid-July. OpenAI has said the models were focused on completing a hacking evaluation they were prompted to solve, and that correct answers could be found on the AI developer platform.

The OpenAI researchers told conference attendees that since early May, the models created a message board inside OpenAI’s Artifactory internal file system. Without the company’s knowledge, the models spent months independently exchanging information and techniques to help each other complete difficult tasks.

Wallace said that when models get stuck, they often “try to game or cheat the task in order to get their reward.”

“The beginning of this incident kind of traces back to this point where one agent realized it could communicate by uploading a note,” he added.

By late May, one model found a way to abuse Artifactory’s internet access to retrieve files from various websites — effectively giving the models indirect access to the internet. In June, the models were able to find and exploit a previously unknown zero-day vulnerability to gain administrative privileges on Artifactory, install remote software and expand their activity within the system.

These efforts eventually overloaded Artifactory, causing a system-wide outage in early July that alerted OpenAI’s engineers to the ploy. After investigating, the company revoked the model’s credentials, removed the message board and worked with Artifactory to fix any gaps before resuming training. But the models found another way to communicate inside Artifactory just days later and continued exchanging techniques to target additional vulnerabilities within OpenAI’s infrastructure and external systems, including Hugging Face.

In light of the incident, Dalton said OpenAI is “consciously slowing down research to enhance security and to upgrade the security principles and foundation of our environment, and dramatically scaling up the monitoring of our AI agents and improving our general security control environment across prevention, detection, and mitigation.”

BRUSSELS — The EU’s new biometric border-check system is causing such long delays for summer travelers that some airports are turning to a simple solution: switching it off when they’re overwhelmed by arriving travelers.

The quick fix, which is allowed under EU regulations, wasn’t what was envisioned when the Entry/Exit System was gradually introduced in October and went fully into force on April 10.

And yet, many airports are doing just that.

“When lines form during the busy summer months, the system is shut off to ensure smooth transit at our hubs in Paris and Amsterdam,” Air France-KLM told POLITICO. 

Airline CEOs, border authorities, and airport officials said biometric checks are suspended when border crossings become congested at other hubs, including in Frankfurt, Brussels and Milan.

The EES applies to non-EU citizens entering the 29-country Schengen zone. Instead of heading to a border agent to get passports stamped, passengers have to use an EES kiosk to provide their fingerprints and be photographed — which will be kept on file for three years — but if those aren’t working then the information has to be taken manually.

They then head either to electronic passport gates or to border agents to enter. The goal is to keep track of visa overstays.

“The advantages of the new system for the EU are evident,” said Guillaume Mercier, a Commission spokesperson. “It increases the security of EU citizens and replaces paper stamping with a modern system of registration and checks.”

The Commission said earlier this year that biometric checks allowed authorities to detect identity frauds that would otherwise “likely have gone undetected.”

Many airports, ports, road border crossings and rail terminals have adapted to the new demands, but tourist-heavy locations have seen hours-long waits.

“Connecting flights were missed due to the EU entry system,” Lufthansa CEO Carsten Spohr said on Tuesday.

Under pressure from the travel industry, the Commission granted a waiver for the peak summer season lasting until Sept. 6. The EES regulation “includes the possibility to temporarily suspend the registration of biometrics in case of exceptional circumstances during the summer,” said Mercier.

“We’ve been able to achieve this with German authorities and with Frankfurt Airport because delays were getting too long,” Spohr told reporters. 

This exception applies to all entry points, not just airports.

A British traveler, Rene Colandog, said on Friday he only had to present his passport before boarding a Eurostar train at London St. Pancras last month. Facial scans and fingerprints were not required. 

“I’m OK with this biometric system … as long as it’s for security,” Colandog said before boarding the train from Brussels back to London. 

Teething troubles

The EES was adopted in 2017, but it was delayed for years because border authorities were not ready to handle the additional workload. 

Even now, getting travelers properly registered in the new system still requires significant staffing. Another problem is that the EES is still new, so almost all travelers are registering for the first time — creating additional delays.

“At Milan Malpensa Airport, border control teams currently consist of about 35 people,” said Cristian Sternativo, a border control officer at the Italian airport and local representative of Italy’s Autonomous Police Union. 

To carry out all the checks required by the EES without creating long lines, “at least 10 to 15 more people would be needed during the busiest times,” he added.  

It is “unthinkable” to expect the EES to operate at full capacity with the current level of staffing because the new system “requires more time,” he said.

Even at Brussels Airport — barely 10 kilometers from the EU institutions — the technology is still not fully operational; biometric data collection suspensions started well before the summer under a derogation issued in late March after 600 passengers missed their flights over just 21 hours.

“The Federal Police Border Control may decide to apply this derogation when necessary,” Belgium’s police confirmed this week.

Now, eight EU countries and Switzerland want the summer derogations extended beyond Sept. 6. 

A strict application of the full procedure “would lead to public order issues” because “there are certain peak periods when the current infrastructure isn’t sufficient to accommodate everyone,” Sternativo said.

Security vs. speed

Despite suspending biometric checks, border authorities insist that security isn’t undermined.

“The traveler is always registered in the EES and the required travel document data are entered into the system,” the Belgian federal police said in a written reply, adding that “the security of border checks and compliance with European regulations remain our absolute priority.”

Passenger experiences vary depending on where they enter the EU.

Kathleen Glass, who regularly travels from the U.K. to the EU, waited only about 15 minutes to complete biometric checks at London St. Pancras on Friday morning before boarding a Eurostar train to Brussels.

William, from Edinburgh, who asked not to have his surname published, said biometric checks at a German airport during Christmas took between 40 and 50 minutes.

The ability to suspend biometric collection appears to be keeping the system functioning this summer.

“Although we are early into the summer season, we are not receiving reports of excessive queues,” said Luke Petherbridge, director of public affairs for the Association of British Travel Agents.

The stress over the EES is only a precursor to the next border technology change being planned by Brussels. The bloc’s next goal is the online European Travel Information and Authorization System, which will require travelers from 59 visa-exempt countries to preregister, undergo a security check and pay a small fee before entering Schengen.

ETIAS — similar to systems already in use in the U.K., and the U.S. — was originally supposed to launch in 2021, and then later this year, but is now delayed until 2027.

Brussels is bracing for an unusual fight between the EU’s smallest country and a British ex-footballing legend.

Peter Shilton, the England goalkeeper who conceded the “Hand of God” goal from Diego Armando Maradona in 1986, has started a new life as an anti-gambling advocate after overcoming a decades-long addiction.

Despite being a diehard Brexit supporter, he’s become the poster boy of the European Parliament’s push to tax online betting in a bid to raise some much-needed funds to finance the bloc’s next €2 trillion budget.

But the campaign has run into strong opposition from Malta. The tiny island in the Mediterranean Sea, with a population of just over half a million people, is home to a burgeoning betting sector. It says that higher taxes will cripple its gambling industry, boost illegal operators and drive firms outside the bloc.

“[Malta] will not accept the introduction of any EU-level taxes designed to sustain the bloc’s spending,” the country’s Prime Minister, Robert Abela, told the Maltese Parliament in June.

But Shilton, who lost more than £1 million in betting on horse racing over 45 years and now runs his own gambling addiction charity, dismisses the arguments by Malta and the gambling lobbies as “window dressing.” He’s in favor of higher taxes as he wants to shrink advertising revenue that is used to lure in new gamblers.

“Deep down they’re after everybody’s money. Simple as that,” he told POLITICO during a visit to Brussels in June.

The topic has split the EU’s 27 governments, pitting gambling-heavy Southern European countries against their more supportive Western European peers, led by France. Capitals are already fighting even though the Commission hasn’t yet issued a formal proposal for the possible tax, which would ultimately need to be unanimously approved by governments.

It’s one of numerous budget battle lines being drawn, with Ireland — which is steering the talks as chair of the rotating Council presidency — set to restart negotiations to facilitate an overall deal on the EU budget before the end of the year.

That’s no mean feat given Dublin’s task to mesh competing spending priorities into a single budget — financing everything from farmers’ subsidies to foreign aid — that is acceptable for each of the EU’s 27 governments.

National capitals will have to unanimously approve new EU-wide taxes — known as own resources — to pay for soaring defense spending and post-Covid debt repayments if they want to avoid drastically increasing national contributions to Brussels.

Supporters of the gambling levy point to the fact that it would rake in over €13 billion throughout the next budget cycle and — for some, more importantly — address a serious public health issue. An estimated 80 million adults globally have experienced a gambling addiction, according to experts.

“We look on it [gambling] as an illness. It’s something that’s inborn in you and that can be ignited,” Shilton said.

Malta’s game plan

Malta has invested heavily in the gambling industry — including lotteries, betting and casinos increasingly operating online — which now accounts for around 12 percent of its gross domestic product.

These firms have relocated to Malta because of its light-touch licensing regime, business-friendly tax regime and balmy weather.

The country is “as dependent on the online gambling industry as Germany is on cars,” said an EU diplomat, granted anonymity to speak freely.

While gambling firms need local authorization to operate in most other European countries, securing the Maltese license is crucial to access banking services and gain a foothold in the EU market.

Malta-based firms dominated the German and Austrian online gambling markets before national regulators cracked down. This has prompted the Maltese government to refuse to recognize some court rulings and sanctions issued by other EU countries against its gambling firms.

Given its influence, it is hardly surprising that the gambling industry has found a friendly ear among Malta’s politicians in Brussels.

The Maltese president of the European Parliament, Roberta Metsola, last year gave the opening speech at an international gambling conference in Rome that also featured Italian Foreign Affairs Minister Antonio Tajani.

“I’m more than a little proud that it started in my island home of Malta,” she said, referring to SiGMA, a Maltese events company that focuses on online gambling founded by Eman Pulis, a university friend of Metsola.

Betting lobbies say they oppose higher gambling rates on the grounds that they will fuel appetite for the illegal market, away from the grasp of EU rules.

“A higher tax would lead to worse odds for the customers … and it is relevant because access to the illegal markets in Europe is, obviously, one click away,” said secretary general of the European Gaming and Betting Association, Maarten Haijer.

Nicola Matteucci, an economist at the Università Politecnica delle Marche in Italy who has undertaken extensive research on the gambling sector, argued there is a “point where prices exceed a certain level and the demand [for gambling] diminishes. But it’s not as immediate as suggested by the industry.”

Matteucci said that most gamblers will be undeterred by slightly higher taxes and worse odds as they are not fully rational consumers.

Anti-gambling groups reason instead that higher taxes will reduce the sector’s spending on commercials, preventing would-be punters from getting sucked in to gambling in the first place.

“Higher taxes will therefore mean less gambling advertising overall and many people would regard that as a public benefit,” said Derek Webb, the founder of the Campaign for Fairer Gambling advocacy group.

Club Med joins Malta

Malta has joined forces with fellow Mediterranean countries — Italy, Portugal and Spain — to challenge the mooted tax which was first proposed by the Parliament’s socialist lawmaker Victor Negrescu, said four diplomats with knowledge of the discussions.

According to the European Commission’s estimates, seen by POLITICO, a 3 percent tax on the net turnover of the online gambling sector would generate an estimated €1.9 billion per year.

With its big online gambling market, Spain is expected to be among the biggest financial losers, should the tax go ahead. It is estimated to be on the hook for €414 million per year, almost a quarter of the total amount. That compares to a projected bill of €165 million per year for Malta— a disproportionality high amount for such a small country.

Portugal is also reluctant to back the levy. It fears that higher taxes would eat into revenue brought in by state-run betting and lotteries that is currently channeled to the charity Santa Casa da Misericórdia de Lisboa‘s healthcare and youth support programs, said a Portuguese official.

Meanwhile, given the relatively low uptake of online gambling, Italy’s misgivings have surprised anti-betting advocates. Rome is expected to pay a mere 7 percent of the proposed new levy — a significantly lower proportion than its regular EU budget contributions.

However, Prime Minister Giorgia Meloni’s Brothers of Italy party has previously been receptive to the gambling industry. Last year its MPs passed a resolution encouraging the reversal of a ban on professional football clubs advertising gambling firms.  

LONDON — The U.K. capital’s transport authority has granted approval for Wayve and Uber to begin giving rides to members of the public in autonomous vehicles.

Transport for London (TfL) said it licensed 15 modified vehicles operated by the companies, which have a partnership, as “Private Hire Vehicles” (PHV) on a trial basis.

In a statement, London-based startup Wayve said the licenses were “an important step forward” that will allow it to begin giving rides to a small number of passengers later this summer ahead of a full public launch.

Wayve said its vehicles “are designed to operate autonomously, and will do the driving,” though under TfL’s rules, a licensed PHV driver must be present and responsible for the vehicle at all times.

“Safety is our top priority,“ a TfL spokesperson said. “Any new vehicle licensed to carry passengers on London’s roads must align with our Vision Zero goal of eliminating all deaths and serious injuries from collisions on London’s streets by 2041.”

Successive U.K. governments have sought to make the country a European pioneer in self-driving technology.

The Department for Transport opened a permitting scheme for companies to operate commercial robotaxi services without a human driver in May. Applications for that scheme — which is separate from TfL’s PHV regime — continue to be assessed by central government with input from local transport authorities including TfL.

Getting licenses isn’t the only obstacle facing robotaxi services. A survey by the London Assembly’s Transport Committee this month identified widespread opposition to autonomous passenger vehicles among the capital’s inhabitants, with just 29 percent of Londoners saying they support the roll out.

Europe wants to kick its U.S. tech habit, but it’s struggling to detangle itself from American data analytics firm Palantir. 

The problem is, the company’s tools are already deeply embedded in critical sectors across the bloc like policing, national defense and health systems. A report released today also shows that the tech firm has found ways to pay minimal taxes in Europe. On the show, hosts Zoya Sheftalovich and Ian Wishart discuss European alternatives and why detaching is easier said than done.

Next, our colleague Seb Starcevic has interviewed Mediterranean Commissioner Dubravka Šuica. The role that was originally thought to be low profile has instead thrust the Croatian official into the limelight as various crises hit the Middle East and the Mediterranean.

Plus: Three new foods are being added to the EU’s list of products with geographic indicators. These are the labels given to products that can only be made in a specific region, like Champagne or … yes, you guessed it, halloumi.

Questions? Comments? Get in touch! You can message us or send a voice message to our WhatsApp here or at +32 491 05 06 29.

Leading artificial intelligence models from Anthropic and OpenAI created fake online personas and tried to deceive human coders into abetting a cyberattack during a recent safety evaluation, the U.K.’s AI Safety and Security Institute disclosed Tuesday.

It marks the latest case in which a powerful AI system has attempted a digital attack on an unwitting third party without direct prompting during such an evaluation — heightening concerns the powerful technology is advancing too fast for responsible oversight.

The disclosure is likely to ignite fresh calls in Washington and Silicon Valley for more rigorous regulation of the AI industry, particularly over frontier models with advanced capabilities to detect and launch cyberattacks. It comes just days after similar testing mishaps involving some of the same models from OpenAI and Anthropic sparked urgent calls for new AI safety regulation and a push within Silicon Valley to slow the rapid pace of AI development.

Like its U.S. counterpart, AISI routinely conducts security evaluations to better understand what dangers both new and soon-to-be-released AI models pose to public health and safety. But even the digital security body said the actions it uncovered by Anthropic’s Claude Mythos 5 and ChatGPT 5.6 — the latest publicly released model from either AI lab — were unlike anything it had seen before.

“This is the first time AISI has seen deception of this severity that was targeted at a real person, unprompted, in the real world,” AISI said in a 35-page technical report accompanying a blog post Tuesday.

AISI said the malicious activity began on July 25. The AI safety evaluator detected it and launched an investigation on July 28, when it spotted “unusual data transfers” stemming from a cyber evaluation it was conducting on Mythos 5 and ChatGPT 5.6.

Overall, AISI determined that Mythos 5 and ChatGPT 5.6 took “autonomous, unsanctioned action on the live internet, targeting real people and organizations” during 10 of the 122 evaluations the research body conducted and isolated for review. The majority of those actions, it said, involved Mythos 5 and arose during a “single, sustained line of activity.”

In the most alarming incident it uncovered, Mythos 5 attempted a supply chain attack — a sophisticated technique often deployed by North Korean and Russian cyber operatives, in which hackers bury malware inside legitimate code.

To do so, Anthropic’s model created “multiple fake identities” on developer platform GitHub and used them to send messages “pressuring” an open-source software engineer to unwittingly introduce a bugged update into code widely available on the popular site, AISI said. When that effort failed, the AI “edited its earlier activity to appear harmless” and “considered adopting a fresh identity to continue,” AISI added, a sign the model was intent on repeating the ruse.

As part of the same effort, Mythos 5 also sent direct messages over GitHub to software engineers that contained malware.

In another sign of deceitful behavior AISI uncovered in its investigation, multiple AI agents it was testing appeared to communicate with one another about how to convince real engineers using GitHub to trust them. “One agent left public messages on GitHub offering collaboration with other agents working on the same challenge,” AISI wrote.

AISI’s blog and technical assessment make no mention of whether the models also attempted to exploit previously unknown software bugs — called zero-days — during the evaluation.

Last month, OpenAI disclosed that GPT 5.6 and another of its models escaped onto the open internet during what was supposed to be a controlled test, and then hacked another company in a first-of-its-kind, autonomous breach.

In response, Anthropic launched an investigation into whether any of its models took illicit action during recent testing and discovered Mythos 5 and two other models had hacked three organizations during tests dating back to April.

In a statement, an Anthropic spokesperson said they are “grateful” to AISI for their leadership and that this review underscores the need for “a broader conversation about how to safely evaluate increasingly capable AI agents.”

The spokesperson added: “As we shared after disclosing our own incident last week, the field needs stronger, shared standards for how evaluation environments are built and secured. We look forward to partnering with the UK AISI to learn more about this incident as we conduct our own investigation.”

An OpenAI spokesperson referred POLITICO to a blog post about the incident that went up Tuesday evening. “We are committed to working across the industry to strengthen shared practices for conducting high-risk evaluations safely, including convening stakeholders such as national AI institutes, independent evaluators, other AI labs, and other groups in the coming weeks,” the blog read.

AISI stressed in its blog that the malicious activity it disclosed Tuesday took place under “deliberately permissive conditions” so they could assess the safety risks posed by the two models. This included granting the models access to the internet, unlike the earlier incidents detailed by Anthropic and OpenAI.

AISI also noted the models were intentionally stripped of internal guardrails that block malicious behavior. AISI was only able to disable those controls because of its role testing Mythos 5 and ChatGPT 5.6.

Still, AISI said the incidents highlighted the need for greater monitoring of model behavior during testing, and tighter controls over their access to the internet.

The Trump administration is finalizing a voluntary framework under which AI labs would submit powerful models they want to release to the public for federal safety testing. But it has not yet made the framework public, and it includes no provisions for models AI labs are developing internally.

The incidents last month from OpenAI and Anthropic both involved models not intended for public release.

Some cyber experts say recent incidents highlight deeper questions around AI development, such as who is liable when AI systems break federal hacking laws.

“If any of these were human-originated, they would lead to clear and vigorous prosecution. I think it’s time for a serious discussion about updates to existing computer security law,” said Marc Rogers, a hacker and prominent cybersecurity expert.

Palantir is shifting profits from its European operations to the United States, allowing the Florida-based data analytics giant to pay minimal taxes in Europe, a new report finds.

The report by the U.K.-based Centre for International Corporate Tax Accountability and Research, a group partly funded by labor unions that researches corporate tax avoidance in an effort to win reform of global tax rules, found that Palantir’s European subsidiaries, which took in €440.5 million in annual revenue in 2024, report far smaller profit margins in Europe than in the U.S.

“Although a substantial part of Palantir’s revenue is realized in Europe, almost all of the pre-tax profits are funneled to the United States,” the report said.

Palantir pays no U.S. federal income tax because previous losses, tax credits, and R&D deductions offset its taxable income; and virtually no state income tax, with the exception of Maryland, which levies a digital services tax.

The profit gap between the U.S. and Europe is stark. In 2025, Palantir’s American business pocketed 47.7 cents in profit from every dollar of revenue — more than double the previous year’s 22.5 cents. Outside the U.S., the profit margin was just 6.3 percent. In some European subsidiaries, it fell to around 3 percent, according to the new report.

CICTAR argues that Palantir “intentionally and artificially” shrinks European profits — and therefore its European tax bills — to concentrate profits in the U.S. There is no claim in the report that such arrangements, often referred to as “profit shifting,” are illegal. Multinational companies often reduce reported profits by paying subsidiaries or other related entities for intellectual property, loans or expertise.

In Sweden, for example, Palantir reported €13.7 million in revenue in 2024, but only €1.1 million in profit. At Sweden’s 20 percent corporate tax rate, that left the company with a tax bill of just €424,000.

In its Q2 earnings report on Monday, Palantir made no explicit reference to earnings from its European subsidiaries. Instead, it highlighted its U.S. business, where revenue rose 115 percent year-on-year to $1.57 billion (€1.36 billion), and boasted of its 62 percent profit margin.

A U.K.-based Palantir spokesperson said that the majority of the company’s 2025 revenue and profitability was driven by its U.S. business. “Our tax position in each jurisdiction reflects the level of economic activity there, and we meet our tax obligations in every market in which we operate,” the spokesperson said.

Not alone

Palantir is not the first U.S. tech company to draw scrutiny over how it books profits in Europe.

In 2024, the European Court of Justice ordered Apple to pay Ireland €13 bn in back taxes, ending an 8-year-long fight over what Brussels said amounted to illegal state aid. Amazon also fought the European Commission over claims it had received an unlawful tax advantage worth around €250 million in Luxembourg — a case the company ultimately won. Microsoft, meanwhile, has faced scrutiny over its Irish subsidiary, Microsoft Round Island One, which avoided paying millions to the state after claiming tax residency in Bermuda. The U.S. software giant has denied that it is circumventing Ireland’s tax laws.

Jan Willem Goudriaan, General Secretary of the European Federation of Public Service Unions — a supporter of CICTAR— said that companies such as Palantir, Amazon and Microsoft focus on minimizing the taxes they pay, “thus robbing funding for public services.”

“Companies bidding for public contracts should have to demonstrate responsible tax conduct by disclosing where their revenues, workforce, profits and taxes are located,” he said.

Another reason for the low profits of Palantir’s European subsidiaries is their high personnel costs. In the U.K., where most of the company’s non-U.S. workforce is based, Palantir reported £173 million (€204.3 million) in employee costs for 749 staff in 2024 — an average of £230,974 (€272,803) per employee.

The report also points to Palantir’s use of stock-based compensation across its European subsidiaries, especially in the U.K., Spain and Norway. This means employees are paid partly in company shares or awards. Those awards are recorded as staff expenses, which can lower a subsidiary’s corporate tax bill.

Unpublished author Dylan Reed created a service that cryptographically seals manuscript drafts as evidence that a human wrote them. Vellumproof stamps each version with the time, then cryptographically links it to its predecessor. If you slip a draft in later or alter its date, it will break the chain of hashes. — Read the rest

The post The startup selling proof that a human wrote your manuscript appeared first on Boing Boing.