You are not who you think you are. You’re not even what you think you are.

Let's start with a story that should be at least somewhat familiar.

One upon a time, when we all lived in Africa and no one lived anywhere else, a people we now call Homo erectus struck into the world. Over the next million years, they spread through the Middle East, Europe, and Asia. They were tall. They were clever. They made sophisticated tools, found their way to isolated islands miles over the horizon, and created the oldest art that came from human hands.

Near the end of that million years, another species, Homo heidelbergensis, developed from an isolated group of H. erectus. We don't know quite where this happened, but in the following half a million years, this species followed their predecessors into Europe, Asia, and the Middle East.

From H. heidelbergensis, other species developed. Around 400,000 years ago, there was Homo neanderthalensis and another species that is alternatively called Homo longi or Denisovans*. (Some analysis had suggested that there was another species in between: Homo antecessor, which is known largely from remains found in Western Europe, but this now looks more like just another offshoot in our bushy family tree.)

Finally, in Africa, another group of H. heidelbergensis produced individuals that appear to be the earliest known examples of Homo sapiens. The oldest we know of came from a place called Jebel Irhoud in Morocco. They are thought to date from around 315,000 years ago.

There were a few false starts, but finally, around 50,000 years ago, H. sapiens successfully followed their relatives out of Africa and into the Middle East, Europe, and Asia. And from there ...

But this is where the story gets strange. Because we–are not those people. You are not those people.

When those last-out-of-Africa hominin strangers, H. sapiens, finally stumbled onto the world stage, they were moving into a place that was practically rife with other humans. We're so used to a world when "human" means just one thing. Over the last three million years, that's seldom been the case.

Those first wandering H. sapiens had a lot of neighbors. Our old ancestor H. erectus had left behind several descendants, such as Homo luzonensis in the Philippines and the diminutive Homo floresiensis in Indonesia. There were probably others we still haven't found. There may even have been groups of plain old H. erectus itself still hanging around to greet the first H. sapiens that came to their neighborhoods.

In both the Middle East and Europe, there was groups of Neanderthals. In Asia, there were Denisovans. We know they met, not just because some sites have been found that share tools, bones, or burials from more than one species. We know, because they are still with us.

You're probably familiar with the fact that many modern people got between 1% and 4% of their genes from Neanderthals. This is so well known that most commercial DNA tests will happily report on your Neandertal-ity along with telling you whether your ancestors are from Denmark or Dalmatia. If it happens that your ancestors all came from Sub-Saharan Africa, your percentage of Neanderthal DNA is much lower. However, there is still a trace, likely resulting from sharing genes with some H. sapiens who had wandered into Neanderthal territory, only to return some generations later.

After two centuries of arguing about it, a pretty good case can be made that, far from murdering or displacing Neanderthals, H. sapiens simply absorbed them. From what we've been able to find, it appears that Neanderthals never existed in great numbers. The percentage of Neanderthal genes we see today may simply reflect the relative sizes of the populations when the two came together.

As multiple scientists and authors have pointed out, there is more Neanderthal DNA today than at any point in history. That's a strange kind of extinction.

From this, you might believe that modern people today take their genes largely from those H. sapiens who developed around Jebel Irhoud, with a dash of Neandertal thrown in. But this story isn't that simple.

Those other descendants of H. heidelbergensis, the ones we call Denisovans (which are probably, probably a part of H. longi) also got into the big ... let's just say "group hug." Many people carry only a trace amount of Denisovan DNA, but if your ancestors came from some areas of Asia or South Pacific islands, you may be carrying as much as 4% of Denisovan DNA with you today.

We know these people from only a handful of fossils, but with numbers like that, they must have been pretty numerous — and friendly — when H. sapiens arrived at their camp fires.

So, humans on the planet today are a mixture of H. sapiens, H. neanderthalensis, and H. longi. In a world where people so often worry about races, we are, all of us, hybrids that span multiple species.

But wait! That's not the big news. All of this is just the build up to a research paper that came out last month in the journal Science.

A group of scientists, from the University of California Berkeley and Johns Hopkins University developed a system they named TRACE (TRacking Archaic Contributions via ARG Estimation ... because it's not science with an acronym) that "uses features of ancestral recombination graphs to identify archaic ancestry" in the DNA of modern humans.

Using examples of complete genomes from over 500 people from around the planet, they went fishing through the code to see which genes were shared, which had diverged over time, and which came from ancestors with "profound" differences from modern humans.

With this tool, they confirmed previous research identifying DNA sequences gleaned from Neanderthals and Denisovans (though their numbers for the percentage of each were a bit more conservative than previous studies). However, that was only the warm up for the big discovery.

In our DNA are at least two "ghost lineages;" fragments of human species who are our ancestors, but whom we have never met.

One of these ghosts lies far back, perhaps as far as two million years. This is the echo of a shared union that doesn't just precede H. sapiens, it may have occurred before H. erectus. (In fact, though the authors don't suggest this, the date opens the possibility that this cross played a large role in creating H. erectus from the base clay of its known ancestor, Homo habilis, and an unknown partner.)

The other ghost in our cells is much more recent. Sometime before H. sapiens left Africa, there appears to have been interbreeding with another species, perhaps an unknown descendant of H. heidelbergensis, because the genes we took from them have a resemblance to those of Neanderthals and Denisovans. It may have been a distinct population of H. heidelbergensis itself, since the youngest specimens of this species overlap with the oldest known H. sapiens by around 15,000 years. If that sounds like our ancestors mated with their own parents, no it's just ... Hey, what's a little genetic incest between species?

Most humans alive today appear to get from 0.2% to 1% of their genomes from this mystery union.

These aren't the only ghosts.

In individuals from Oceania, TRACE found an average of 0.73% Neanderthal, 0.66% Denisovan, and 0.33% ghost ancestry. Those numbers for the first two are lower than previous studies, but the TRACE data also shows something else interesting—hints of another unknown ancient hominin dating back to 1.8 million years ago. Because these traces were highest in the groups that had more Denisovan ancestors, the researchers suspect that it was the Denisovans who carried these "super-archaic" genes.

We find a significantly higher proportion of super-deep lineages in Denisovan than in Neanderthal segments. Simulations show that this pattern is inconsistent with a model lacking super-archaic introgression but is recapitulated by a model including super-archaic introgression into Denisovans. Furthermore, super-archaic fragments within introgressed Denisovan segments exhibit distinct genetic features that differentiate them from the background Denisovan ancestry, including much deeper coalescence times with modern humans, longer genomic lengths than mean ILS length from Neanderthal deep lineages, and low affinity to both Neanderthal and Denisovan reference genomes.

How Denisovans might have preserved these ancient genetic traces if they branched from H. heidelbergensis at about the same time as Neandertals isn't clear. All of us probably share in this ancient mixing, but because our percentage of Denisovan DNA is so low, for most of us it's barely a genetic whisper.

We've known for some time that the "march of progress" image from the 1960s—the one that shows a chimp-like ancestor on the left following by a series of brutish, slouching intermediaries leading to a modern (European) man on the right–is horribly wrong. In fact, paleontologists knew that idea was wrong before it was ever published. But it's only in the last two decades that we've learned just how wrong it is.

The story of modern humans and our hominin ancestors is complex. The tree of our ancestors is filled with limbs that went nowhere, and branches that confound expectations. And, unlike most actual trees, some of those branches seem to have merged as well as split.

We who survive are hybrid humans, carrying in every cell the story of ancient travels and forgotten people. Those H. sapiens at Jebel Irhoud are some of our ancestors, but they're far from the only ones. We each contain multitudes. None of us is "pure," and we should be glad of it.

There's one more twist to this story.

In a 2025 paper in Nature, a group of scientists from the University of Cambridge used their own tool (COBRAA) to sift the human genome for markers that indicated major events and population size changes.

What they found was that humans today are not descended from one group of ancestors, but two. These groups became separated from each other around 1.5 million years ago, possibly around the time some H. erectus migrated out of Africa and went off into the broader world. Soon after they split, one of these groups – "Population A" fell on hard times. It faced a population bottleneck that left it with only a few survivors, limiting their genetic diversity. Still, it was Population A that eventually produced H. heidelbergensis, Neandertals, and Denisovans.

Meanwhile, Population B flourished. They kept their genetic diversity and their population was more stable.

Whatever the reason for the separation, the genetic markers indicate that around 300,000 years ago, these two groups met up again. Whoever they were by then. Whatever they were. They came together with all the differences they had accumulated over a million years of being apart. After all their wandering and all they had seen.

We're not Population A or Population B. We're both.

Using cobraa, we present evidence for an extended period of structure in the history of all modern humans, in which two ancestral populations that diverged ~1.5 million years ago came together in an admixture event ~300 thousand years ago, in a ratio of ~80:20%.

Those people who appeared in Morocco came directly from this merger of two long-separated groups. When they left Africa again, they gathered into themselves all the scattered forms of human that they met. We're all of them now. We're all of them. We're all the eggs, and the basket.

You are 80% the people of Population A, who went through a terrible time, but survived. You are 20% the people of Population B, who flourished in their time, before welcoming their long-long-long-lost relations.

You're part Neanderthal, part Denisovan, and part ghost. You're 100% human. So are we all.

Today's links

• IP can't save you from AI: Property rights cannot substitute for labor rights and privacy rights.
• Hey look at this: Delights to delectate.
• Object permanence: Flying Brits v brown people; Probability neglect; Law v "enhanced patdowns"; Onion says Brits love paywalls; "Hench"; "Lessons in Magic and Disaster."
• Upcoming appearances: Sydney, Melbourne, Brighton, London, South Bend.
• Recent appearances: Where I've been.
• Latest books: You keep readin' em, I'll keep writin' 'em.
• Upcoming books: Like I said, I'll keep writin' 'em.
• Colophon: All the rest.

IP can't save you from AI (permalink)

You don't have to believe that AI "art" is any good (I don't), nor do you have to believe that AI "art" can be any good (I don't) to understand that the reason that the capital markets are putting trillions into AI is that they believe they can fire workers of every kind and replace them with AI:

https://pluralistic.net/2025/03/18/asbestos-in-the-walls/#government-by-spicy-autocomplete

I'm an artist and a worker. I want to protect my labor interests. So do my peers from across the "creative industries." But a sizable group of my peers think the way we're going to protect our interests is by expanding copyright so that it's unambiguously illegal to scrape the internet, analyze the files retrieved by those scrapers, and publish that analysis (a process more familiarly known as "training AI"):

https://pluralistic.net/2023/09/17/how-to-think-about-scraping/

This is a losing strategy. First, because banning scraping, or requiring permission to count the elements in creative works, or demanding a license to publish collections of facts about copyrighted works will inflict enormous collateral damage on a wide variety of socially beneficial activities. From the OED to search engines to the Internet Archive, so many beneficial activities rely on the fact that copyright permits unlicensed collection and analysis of every copyrighted work as a single, massive corpus, and copyright allows the publication of that analysis without permission from the creators of the works it analyzes.

A lot of people who are (rightfully) very angry about AI dispute this. They believe that they can craft an "AI training" law that would ban scraping, analysis and publication when these activities are part of AI training, but not when they're undertaken for a benign purpose. I am very, very skeptical of this. After 25 years of watching internet policy go badly awry, to the great detriment of workers of all kinds and everyday users, it is my professional, considered opinion that drafting a statute that only stops these "bad" activities is much, much harder than these people think, and may actually be impossible.

I think some artists advocating for a copyright-based solution to AI's war on labor understand this and have decided that they're willing to catch a lot of dolphins in these legal tuna-nets they're hoping to get from Congress. I get that: there are always trade-offs, and the perfect can't be the enemy of the good.

But I think they're making the wrong trade-off, and not just because I value archives, accountability corpuses, large-scale linguistic research and search engines. I think they're making the wrong trade-off because copyright will not protect their livelihoods from AI-based wage erosion.

Here's why: the theory of copyright as an "artist's right" is premised on the idea that we artists get these exclusive rights, which we use in our bargaining with media companies and other intermediaries. It's a (pseudo) property right, and it's sub-licensable. Just as an entrepreneur might get the contract to supply catering for a sports stadium and then parcel out the pretzel stand, beer bar, and pizza concessions to subcontractors, we're meant to sell our English rights, foreign language rights, graphic novel rights, film rights, audio rights, (and so on) to a variety of media companies.

To bargain successfully, it's not only necessary for you to have something valuable to trade: you also need to have leverage. You need to have options. The other side has to believe that if they lowball you, you will go do a deal elsewhere.

This is where copyright fails to serve creative workers. Even at the best of times, the world naturally produces an oversupply of would-be professional artists, and a sufficiency of the talented to fill most of the workaday niches in our field. Even exceptional artists – and exceptional works of art – are often commercial flops, for reasons that aren't always well understood (though sometimes it's a self-fulfilling prophecy, where a media company buys the rights and then loses confidence in the work and does not exert itself in the marketing of the work).

These are not the best of times. Decades of lax antitrust enforcement has boiled the "creative industries" down to 5 publishers, 4 studios, 3 labels, 2 app stores, and one company that's in charge of all the ebooks and audiobooks.

Since the 1976 Copyright Act, Congress has acted time and again to broaden copyright. Today's copyright lasts longer, restricts more uses, extends to more kinds of works, and carries stiffer statutory penalties for infringement ($150,000 per download!). The media companies we creative workers bargain with are larger, richer and more profitable than at any time in history – and we are poorer. The share of those massive profits that ends up in our pocket is lower than ever – and we don't just get smaller slices of that larger pie, those slices are smaller than the slices we used to get, when the pie was much smaller. The rising tide of copyright expansion lifted our bosses' boats – even as our dinghies filled with bilge and sank.

How could we get so much more to bargain with, only to bargain it all away, for less money than we used to get for a much smaller bundle of rights? Simple: giving us rights did not give us leverage. Giving us more rights without giving us more bargaining power is like giving your bullied schoolkid extra lunch-money. There's no amount of lunch-money that will get that kid fed; but if you keep increasing how much money the kid gets, the bullies will end up so rich that they can afford to run a global campaign demanding that we all think of those poor hungry kids and send them even more lunch money.

Copyright's failure to deliver for creative workers doesn't mean that we're doomed to poverty. Our works are generating record profits for our bosses, and there are plenty of ways to change the "distributional outcomes" (the phrase economists use for "who gets what") in arts/labor policy. In 2022, I co-wrote Chokepoint Capitalism along with the eminent Australian copyright scholar Rebecca Giblin. The whole book is full of these pro-worker arts policies:

https://pluralistic.net/2022/08/21/what-is-chokepoint-capitalism/

Rebecca and I start from the premise that artists are workers, not the small businesses that our bosses insist we see ourselves as. The idea that an artist is an LLC with an MFA fits in very neatly with copyright: you're getting this bundle of exclusive rights from Congress and then you bargain, business-to-business, with other companies out there in the world, selling those rights for the best price you can get. This approach rarely works, and when it does, it works badly. 50 years of more copyright, richer bosses, and poorer artists put the lie to the "LLC with an MFA" approach.

If we're workers, then we derive our power from labor rights. The Writers Guild – the only creative workers in world history to have comprehensively beaten AI in their workplace – won their AI fight with a strike:

https://pluralistic.net/2023/10/01/how-the-writers-guild-sunk-ais-ship/

The Hollywood guilds are able to pursue a limited form of "sectoral bargaining" (where all the workers in a field bargain with all its bosses) called "multi-employer bargaining." Bosses hate sectoral bargaining, and in 1947 they got it banned outright through the Taft-Hartley Act.

Getting other kinds of creative workers into multi-employer bargaining arrangements will be a lot of work – and repealing Taft-Hartley and restoring sectoral bargaining will be even harder. But just because it's hard to do the thing that works, it doesn't follow that we should do the easy thing that doesn't work.

Compared to winning more labor rights, getting more copyright will be easy. That's because our bosses want more copyright. When we demand more copyright, our bosses – the most powerful, profitable media companies in human history, grown rich off our labor – will fight alongside of us.

But media companies don't want to stop AI from depriving us of our wages. Quite the contrary! The whole reason that the Writers Guild had to go on strike was that movie studios – not Openai or Anthropic – wanted to replace them with AI. The same studios that are suing the AI companies for "mass copyright theft" have made it very clear that they want to buy chatbots from those AI companies and use them to erode our wages and thin our ranks. The copyright lawsuits our bosses are waging against the AI companies are intended to force tech companies to pay for licenses before they train their chatbots on our work. But they won't be paying us for those licenses – they'll be paying our bosses.

The AI copyright fight isn't being fought to protect your wages – it's being fought to see whether your lost wages end up in the pockets of a tech boss or a media boss. AI copyright suits are a fight over who's going to get the lion's share when they eat you up for dinner. They're not a way to keep you off the menu.

This becomes more obviously true with each passing day, and this morning, the world got its clearest example of what a poor substitute copyright is for fundamental human rights, like labor rights and privacy rights.

Last year, Spirit Airlines went bankrupt, a casualty of a monopolized aviation sector and Trump's oil price surge. Ever since, vultures have circled its carcass, picking off its assets in a string of auctions conducted by Spirit's bankruptcy trustees. Today, those trustees announced that they had sold all of Spirit's employees' data to Google, for use in AI training:

https://www.axios.com/2026/08/17/google-spirit-airlines-bankruptcy

Every email, every memo, every calendar entry. Oceans of sensitive, personal information, all to be shoveled directly into the bottomless maw of Google's AI training systems. This training data includes messages between colleagues and with outside parties about workers' romantic lives, their health, their family situations. These workers' most private lives will end up as fodder for a Google chatbot.

Now, all of these workers have a copyright in all of that work. Under international copyright treaties and US law, copyright "inheres at the moment of fixation of a work of human creativity." The very instant a worker sets fingers to keyboard and types out a message with even the smallest quantum of creativity, a new copyright springs into existence, giving the copyright holder 90 years' worth of control over it.

But even though every one of those emails and messages and memos was written by a human being working for Spirit, the copyright over those works does not belong to the workers. Every single one of them will have signed an employment agreement that designates their emails and other copyrightable work as "works made for hire," owned by Spirit Airlines, which means that their work is now an asset in Spirit's bankruptcy estate. That's why all that personal information is about to be transferred to a new corporate owner, Google, who can do anything they want with it.

We know how terrible this kind of disclosure will be for workers. In 2001, the criminal enterprise Enron collapsed after the extent of its fraud was revealed. In the ensuing litigation, Enron's bankruptcy overseers decided that it was too expensive to purge the company's email servers of personal information before entering it into evidence. That meant that once the court battles were over, all the Enron employees' emails entered the public domain as part of the court record:

https://en.wikipedia.org/wiki/Enron_Corpus

The "Enron Corpus" is a foundational data-set in modern computer science. Academics analyzed the data to do pioneering work on machine learning and social graph theory, which found its way into the design and operations of social media companies, who learned how to spot and manipulate social connections by studying it.

The Enron Corpus isn't just a data-set, though. It's a privacy catastrophe, full of sensitive personal information that haunts the 158 employees whose correspondence is now permanently afloat upon the internet.

Why was the Enron Corpus so exploitable? Because US labor law does not protect this kind of sensitive information when it is in your employer's hands. In fact, if your boss ends up with a trove of your personal information in the form of emails, calendar entries and files, you will typically be blamed for it: "Why did you use your work computer for personal activities?"

But anthropologists who study computer usage have known for decades that everyone ends up with personal data on their work devices. What's more, this problem is only getting worse, because (thanks to weak labor laws), we're expected to work longer hours and to be on call when we're not at the job, which means that you're often dealing with personal crises after hours from your desk, and dealing with work crises at home from your sofa.

Any fit-for-purpose labor rights regime would recognize that your privacy rights must extend to the data that finds its way onto your boss's computers, even if you put that data there. Any failure to recognize this bedrock fact gives employers free license to plunder and exploit your personal information.

Of course, labor law isn't the only way to protect private information. While labor law should contain explicit, job-related privacy guarantees, privacy law should protect all our privacy (after all, Spirit's servers are also full of emails and messages from Spirit's passengers).

Unfortunately for anyone who ever flew on Spirit – or anyone who worked for them – American privacy law is all but dead. America's last consumer privacy law went into effect in 1988, when the Video Privacy Protection Act made it illegal for video-store clerks to disclose your VHS rental records.

Google says it won't use your profile or frequent flier info to train its model, but they haven't made the same promise about the millions of messages that passengers exchanged with the airline. Google has also promised to use "de-identification" algorithms to purge the Spirit customer, supplier and employee data of personal information. But "de-identification" is a pipe-dream, widely understood by security experts as a form of wishful thinking by companies that want to exploit your personal information while still insisting that they aren't violating your privacy. In reality, "de-identified" data is always vulnerable to "re-identification" attacks:

https://pluralistic.net/2021/04/30/dox-the-world/#experian

The collapse of privacy and labor rights in post-Reagan America and the mass expansion of copyright over the same period are part of the same phenomenon, aspects of two generations' worth of policies designed to benefit capital at the expense of workers, and corporations at the expense of consumers.

As consumers, we're told to substitute shopping for legal rights: if a corporation wrongs you, it's easier and quicker to "vote with your wallet" than it is to sue them or ask the government to intervene. Substituting shopping for politics has been a total failure. Shopping your way out of a monopoly is like recycling your way out of a wildfire:

https://pluralistic.net/2026/05/21/purity-culture/#stop-fucking-that-chicken

As creative workers we were told to stop thinking of ourselves as workers altogether, to become small businesses, and to use the LLC With an MFA method to bargain our way out of exploitative arrangements. This, too, has been a failure:

https://pluralistic.net/2026/03/03/its-a-trap/#inheres-at-the-moment-of-fixation

The sale of Spirit's data to Google for AI training shows us that privacy and labor rights are indispensable. We can't substitute market mechanisms like comparison shopping or individual contract negotiations for broad, systemic, inalienable rights backstopped by law.

By demanding the copyright our bosses love, we're seeking the right to be angry about AI, even as the AI companies and our bosses cut deals to train chatbots with our work, which they will use to attack our livelihoods.

Once we stop pretending to be small businesses, once we abandon the fantasy of LLCs with MFAs, we can join with every worker in every industry in demanding sectoral bargaining; and with every consumer in demanding privacy rights. Winning privacy and labor struggles means more than the right to be angry about AI – that's the right to do something about it.

Hey look at this (permalink)

• Nautilus https://nautilus.plotter.cc/

• Zoomers don't know what Usenet is https://tchotchke.substack.com/p/zoomers-dont-know-what-usenet-is

• When the Shortage is the Strategy https://nooneshappy.com/article/when-the-shortage-is-the-strategy/

• Can Canada function without American tech? https://www.youtube.com/watch?v=WwoL6OGk_2Y

• Elon Musk made flying even worse so Palantir could profit https://www.theverge.com/transportation/981194/faa-air-traffic-elon-musk-peter-thiel-palantir

Object permanence (permalink)

#25yrsago IP and scientific publishing https://web.archive.org/web/20011001203058/http://www.abc.net.au/rn/talks/bbing/stories/s345514.htm

#20yrsago British air travelers kick brown “terrorists” off their planes https://web.archive.org/web/20060823104858/http://www.dailymail.co.uk/pages/live/articles/news/news.html?in_article_id=401419&in_page_id=1770&ico=Homepage&icl=TabModule&icc=NEWS&ct=5

#15yrsago “Probability neglect”: why policy-makers are constitutionally incapable of formulating evidence-based anti-terrorism policy https://web.archive.org/web/20111015040753/https://opim.wharton.upenn.edu/risk/library/J2011OBHDP_APM,AT,HK_PolicymakersDilemma.pdf

#15yrsago TSA can’t explain why “enhanced patdowns” are legal https://web.archive.org/web/20151203033820/http://flyingwithfish.boardingarea.com/2011/08/18/the-legality-of-the-tsas-enhanced-pat-down-authority/

#15yrsago The Onion: We did a paywall because British people like paying for the Web https://web.archive.org/web/20110911175335/http://www.avclub.com/articles/about-the-onions-new-paid-content-system,60129/

#5yrsago Hench https://pluralistic.net/2021/08/19/failure-cascades/#natalie-zina-walschots

#5yrsago Machine learning's crumbling foundations https://pluralistic.net/2021/08/19/failure-cascades/#dirty-data

#1yrago Charlie Jane Anders' "Lessons in Magic and Disaster" https://pluralistic.net/2025/08/19/revenge-magic/#liminal-spaces

Upcoming appearances (permalink)

• Sydney: The Festival of Dangerous Ideas, Aug 23-24
  https://festivalofdangerousideas.com/program/

• Melbourne: Enshittification at the Wheeler Centre, Aug 25
  https://www.wheelercentre.com/events-tickets/season-2026/cory-doctorow-enshittification

• London: AI and the Enshittification of the Media, NUJ (Sep 2)
  https://www.nuj.org.uk/learn/ems-event-calendar/ai-and-the-enshitification-of-the-media.html

• Brighton: The Reverse Centaur's Guide to Life After AI with Carole Cadwalladr (Brighton Dome), Sep 8
  https://brightondome.org/whats-on/LSC-cory-doctorow-the-reverse-centaurs-guide-to-life-after-ai/

• London: The Reverse Centaur's Guide to Life After AI with Riley Quinn (Foyle's Picadilly), Sep 9
  https://www.foyles.co.uk/events/enshittification-cory-doctorow-riley-quinn

• South Bend: An Evening With Cory Doctorow (Notre Dame), Oct 6
  https://franco.nd.edu/events/2026/10/06/an-evening-with-cory-doctorow/

• Victoria: Munro's Books (Oct 20)
  https://www.munrobooks.com/events/6113620261020

• Vancouver: BC Policy Solutions Gala, Nov 12
  https://bcpolicy.ca/gala/

Recent appearances (permalink)

• Deflating the AI Bubble (Do Not Pass Go)
  https://www.donotpassgo.ca/p/deflating-the-ai-bubble-with-cory

• Technofeudal Enshittification (Fucking Cancelled)
  https://www.fuckingcancelled.com/p/technofeudal-enshittification-with

• Who The Machine Serves (EFF)
  https://archive.org/details/effecting-change-who-the-machine-serves

• Speculative Fiction for Social Change II (Cool People Who Did Cool Stuff)
  https://pocketcasts.com/podcast/cool-people-who-did-cool-stuff/08cbb840-a6ae-013a-d8aa-0acc26574db2/part-two-cory-doctorow-on-speculative-fiction-for-social-change/937e8800-9404-45a6-b5e3-90ebee2cfaea

• Speculative Fiction for Social Change I (Cool People Who Did Cool Stuff)
  https://pocketcasts.com/podcast/cool-people-who-did-cool-stuff/08cbb840-a6ae-013a-d8aa-0acc26574db2/part-one-cory-doctorow-on-speculative-fiction-for-social-change/15ad467c-0832-44c9-91ea-59defd783dba

Latest books (permalink)

• "The Reverse-Centaur's Guide to AI," a short book about being a better AI critic, Farrar, Straus and Giroux, June 2026
  https://us.macmillan.com/books/9780374621568/thereversecentaursguidetolifeafterai/

• "Canny Valley": A limited edition collection of the collages I create for Pluralistic, self-published, September 2025 https://pluralistic.net/2025/09/04/illustrious/#chairman-bruce

• "Enshittification: Why Everything Suddenly Got Worse and What to Do About It," Farrar, Straus, Giroux, October 7 2025
  https://us.macmillan.com/books/9780374619329/enshittification/

• "Picks and Shovels": a sequel to "Red Team Blues," about the heroic era of the PC, Tor Books (US), Head of Zeus (UK), February 2025 (https://us.macmillan.com/books/9781250865908/picksandshovels).

• "The Bezzle": a sequel to "Red Team Blues," about prison-tech and other grifts, Tor Books (US), Head of Zeus (UK), February 2024 (thebezzle.org).

• "The Lost Cause:" a solarpunk novel of hope in the climate emergency, Tor Books (US), Head of Zeus (UK), November 2023 (http://lost-cause.org).

• "The Internet Con": A nonfiction book about interoperability and Big Tech (Verso) September 2023 (http://seizethemeansofcomputation.org). Signed copies at Book Soup (https://www.booksoup.com/book/9781804291245).

• "Red Team Blues": "A grabby, compulsive thriller that will leave you knowing more about how the world works than you did before." Tor Books http://redteamblues.com.

• "Chokepoint Capitalism: How to Beat Big Tech, Tame Big Content, and Get Artists Paid, with Rebecca Giblin", on how to unrig the markets for creative labor, Beacon Press/Scribe 2022 https://chokepointcapitalism.com

Upcoming books (permalink)

• "The Post-American Internet," a geopolitical sequel of sorts to Enshittification, Farrar, Straus and Giroux, 2027

• "Unauthorized Bread": a middle-grades graphic novel adapted from my novella about refugees, toasters and DRM, FirstSecond, April 20, 2027

• "Enshittification, Why Everything Suddenly Got Worse and What to Do About It" (the graphic novel), Firstsecond, 2027

• "The Memex Method," Farrar, Straus, Giroux, 2027

Colophon (permalink)

Today's top sources:

Currently writing:

• “Once Is Enemy Action,” a science fiction novel about the origins of modern technofascism. Today's words: 585 (6624 total).

• "The Post-American Internet," a sequel to "Enshittification," about the better world the rest of us get to have now that Trump has torched America. Fourth draft completed. Submitted to editor.

• A Little Brother short story about DIY insulin PLANNING

This work – excluding any serialized fiction – is licensed under a Creative Commons Attribution 4.0 license. That means you can use it any way you like, including commercially, provided that you attribute it to me, Cory Doctorow, and include a link to pluralistic.net.

https://creativecommons.org/licenses/by/4.0/

Quotations and images are not included in this license; they are included either under a limitation or exception to copyright, or on the basis of a separate license. Please exercise caution.

How to get Pluralistic:

Blog (no ads, tracking, or data-collection):

Pluralistic.net

Newsletter (no ads, tracking, or data-collection):

https://pluralistic.net/plura-list

Mastodon (no ads, tracking, or data-collection):

https://mamot.fr/@pluralistic

Bluesky (no ads, possible tracking and data-collection):

https://bsky.app/profile/doctorow.pluralistic.net

Medium (no ads, paywalled):

https://doctorow.medium.com/

Tumblr (mass-scale, unrestricted, third-party surveillance and advertising):

https://mostlysignssomeportents.tumblr.com/tagged/pluralistic

"When life gives you SARS, you make sarsaparilla" -Joey "Accordion Guy" DeVilla

READ CAREFULLY: By reading this, you agree, on behalf of your employer, to release me from all obligations and waivers arising from any and all NON-NEGOTIATED agreements, licenses, terms-of-service, shrinkwrap, clickwrap, browsewrap, confidentiality, non-disclosure, non-compete and acceptable use policies ("BOGUS AGREEMENTS") that I have entered into with your employer, its partners, licensors, agents and assigns, in perpetuity, without prejudice to my ongoing rights and privileges. You further represent that you have the authority to release me from any BOGUS AGREEMENTS on behalf of your employer.

ISSN: 3066-764X

X’s algorithm learns what you hate and shows you more of it, according to a new study just published in the Proceedings of the National Academy of Sciences (PNAS). The paper, titled Value misalignment of X’s feed algorithm is a reflection of value tensions in engagement, found that the site’s algorithm prioritized engagement above all else when it generated a user’s For You Page. It also showed that X serves more ragebait to people who say they are Democrats, although the exact reason for that is unclear.

“In 2026 that’s maybe not the most surprising headline ever,” Ziv Epstein, a postdoctoral researcher at Stanford University, and co-author of the paper, told 404 Media. “So we actually dug in a little deeper to figure out why this is actually happening, and it turns out that X's feed algorithm, like a lot of these social media algorithms, is optimized for engagement [but] it turns out that not all types of engagement are considered equally.”

The study’s goal was to understand how a user’s self-professed values system might shape what they see on X. “We recruited a nationally representative sample of N = 715 Americans who are active users of X in September and October 2024, quota matched on ethnicity, gender and partisanship, to install a browser extension to collect their [For You Page] and Following feeds,” the study said.

Epstein said the study was observational and meant to get people asking questions about what they want to see on social media, how their feed is designed, and by whom. “There are these social media algorithms that have enormous amounts of power in our lives, they shape the information that we consume, and we have very little transparency into how they operate and what their implications are,” he said. “And so, we were very interested in trying to understand the particular effects of this particular algorithm, and so I think that has kind of important implications for civil society and just fighting some of the technofeudalistic tendencies of platforms to control these algorithms.”

For the study, researchers collected a “values inventory” of the volunteers using a research tool called the Schwartz Theory of Basic Values. The values inventory in the study is presented as a wheel with 19 points that corresponded to features like “tolerance,” “dominance,” “hedonism,” and “openness to change.” Users also reported their political alignments.

Then researchers watched how users engaged with posts on X and how those posts reflected their self-reported values. “We observe that the inventory of posts from followed accounts reflects users’ self-stated values — but that there is an overall negative correlation (misalignment) between users’ explicit values and the values in content that the algorithm is more likely to amplify,” the study said.

When a user on X sees a post that makes them mad — like a press release from a politician from a political party they don’t like — sometimes they’ll fight about the post in the replies. It doesn’t matter who you follow or what your stated values are, X reads replying as engagement and will send more of the infuriating posts the user’s way.

“When we look at commenting, the act of replying to posts, that's where we actually see some kind of meaningful misalignment between people’s values and the values of the content they’re replying to,” Epstein explained.

Most of the participants liked and reposted content on X and got served more of the same sort of content. Replying was rare, just 6.8% of the interactions according to the study, but had an outsized impact on the algorithm. “Replying is only a fraction of engagement, but there does seem to be some evidence that these algorithms are prioritizing and learning more from this kind of rarer form of engagement,” Epstein said. “So it’s this feedback loop of outrage baiting. The algorithm learns that you get outraged and then continues to serve more content in that direction and that seems to be particularly true of the Democratic users of our study.”

Though this happened across the political spectrum, the study found that ragebaiting occurred more for users that identified themselves as Democrats. “Democrat users confront the abundant value-misaligned content by replying to it, which the algorithm in turn preferentially learns from and continues to feed them,” the study said. “This highlights a core tension with how engagement-maximizing algorithms operate on social media: frictions between users’ stated preferences and their behaviors of reactive confrontation are exploited by engagement-maximizing algorithms to create runaway feedback loops of increasing value misalignment.”

Epstein said he’d need to do more research to find out why X seems to serve ragebait to Democrats more often than Republicans. It could be that there’s more rightwing content on X overall or it could be that Democrats tend to engage with posts they disagree with more often. “There might be some kind of differential effects on information diets there, or it might be something more psychological about how different you know partisan identities are triggering different kinds of actions and reactions, but ultimately I don't want to speculate too much,” he said.

I asked Epstein if he worried that prioritizing “values” in a social media algorithm might lead to more siloed user bases and more echo chambers. “I do think that if we go kind of down this path of thinking through and imagining value line social media feeds, we do have to be very aware of the potentials of value echo chambers, right?” he said. “Where people just, you know, they have the certain values that they have, and all the content they see is just aligned with those values. I think that is a very scary and dark reality.”

But he also said that values are intentional and that thinking about the kind of stuff you want to see on a social media site before you pull up your feed is a positive. “You're pumping the brakes, you're taking a breath, and you're thinking about what you actually really care about. In this world — and I don't have the data for this — but I would speculate that a lot of people actually do care about seeing a diverse set of content and engaging meaningfully across these lines, and you know maybe that isn't a particular value on my 19-dimensional wheel, but this is just a starting point of thinking about our intentions and thinking very deliberately about the kind of information we want to be exposed to, versus the the knee-jerk reaction that we're kind of learning in this very kind of short, shallow attention span, emotion and negative affect-driven model of these very myopic forms of engagement.”

X did not return 404 Media’s request for comment, but Nikita Bier — X’s former head of product — confirmed that the site’s algorithm had at one time been set to favor replies. “This is no longer true,” Bier said in a post on X. “The largest contributor of seeing ragebait was the reply predictor and we were aware that angry replies were causing people to see more of that content. So last month, we gave the reply predictor a 15x boost if it’s a friend’s post — and it reduced ragebait by [an] order of magnitude.”

Update 8/18/26 at 1:30PM: This story has been updated to include a comment from X’s former head of product

How Claude marks AI-generated content is the most self-destructive utterance I’ve ever heard coming out of an AI company. Wow:

When a supported Claude model generates text, it weaves an imperceptible watermark directly into the text itself. You won’t see it, and it doesn’t change the meaning, quality, or readability of Claude’s response.

Because the watermark is part of the text, it will travel with the text when it’s copied and pasted elsewhere, and may persist through some editing. Watermarking will be applied at the model level, which means it will be present no matter which Claude product or surface the text comes from.

Will any writer, of anything, ever want to use any Claude text anywhere now? I’ve been a satisfied Claude customer, and I’m ready to bail completely. And I’m not alone. Because watermarks in text are invisible time bombs ready to scream, “The author didn’t write this! I did!” when some reader or their machine uses the right probe.

I feel bad for all the writers whose work depends on AI composition. For example, Daniel Barkhuff. In Hey Guys, I use AI! And so does everyone, he writes,

My process isn’t complicated. I sit down and write about a page, maybe 450 words. It’s not elegant. It’s not structured. It’s basically a brain dump. Half sentences, ideas that don’t quite connect yet, things I’d say out loud but that look ridiculous when you see them on the screen. It’s intellectual vomiting. A rough sketch of what I’m trying to say.

Then I paste it into AI and say something like, “Hey, give me an essay.”

And it does. (The tell is a paragraph, then a skipped line, then one sentence, preferably short and under 6 words).

I wrote the parenthetical part of the above.

But here’s the thing people misunderstand about that step. The machine isn’t doing the thinking. It isn’t generating the ethics, the argument, or the moral framework. All of that was already there in the messy words I wrote first. What the machine is doing is what editors have always done: taking raw thought and making it readable. AI doesn’t think. It formats.

Wrong on three counts. 

1. Editors edit the work of other writers. They don’t write for other writers. If they do, it’s called ghostwriting.
2. AIs don’t think, any more than they aren’t trained, don’t have information, and don’t learn. They emulate all those verbs, extremely well. That’s why—
3.  It sells AI very short to say it just “formats.”

With that post, Dr. Barkhuff watermarked everything he has written since, and I’ve read him less for it. And I hate to say that because I respect and admire him a great deal. And I’m sure he’s actually a good writer. 

While I understand why the EU requires this kind of shit, it throws monkey wrenches all over the place—because ghostwriting isn’t the only issue here.

John Gruber says,

Anthropic’s (original) support document unambiguously claims that’s what their system will enable. So if that were true, I couldn’t see what was left other than hiding invisible characters within the text.

My error was believing Anthropic that their system wouldn’t adulterate and corrupt the semantics of the text their models generate. That is in fact exactly what they plan to do…

The idea that anything other than my needs should factor into the generation of text for me is patently offensive.

This isn’t just about text one might generate with the intention of passing it off as their own natural work. This isn’t even about LLM proofreading of work written by hand. Anthropic is saying that all new Claude models are going to adulterate every single bit of text longer than 200 tokens (~150 words) they generate, including everything it presents to its users to read. So even in a private conversation between a user and Claude, which will never be read by anyone other than the user, Claude will begin making word choices in the name of marking its output in statistically predictable ways rather than maximizing clarity and precision.

Jeff Jarvis:

I’m angry as a writer.

Anthropic’s method will alter its models’ word choices in a pattern based on a key only it knows, allowing it to detect writing (or editing or translation) by its machine — to an uncertain degree of certainty. The company contends there is no difference in meaning between the otherwise random choices “grey” or “overcast” in, for example, the context of weather. So what’s the issue?

Thus Anthropic declares words fungible, language random, choice meaningless. Not to me. I try to select my words as carefully as I can for a number of considerations: style, tone, rhythm, avoiding repetition, but most of all meaning. You may quibble with my choices, but they are mine, not yours; that’s what makes my writing mine, to communicate what I wish to communicate. “Gray” and “overcast” are not the same, damnit. That’s the reason we have both of them.

So I resent Anthropic et al dismissing the value of a writer’s selection of one word over another for the impact that decision has on expression, on impression — occasionally on art.

We humans might not be able to detect whether the machine or the key is being used. That’s not the point. The selections/changes/alterations made in accordance with the key will affect sense and style on an ever-larger proportion of text entering discourse, and the fact that Anthropic (and the EU) consider this acceptable, unimportant, even trivial has cultural impact on the worth of words. They devalue writing. At the very moment when LLMs commodify writing and writers, this is just another kick to the kidneys for a challenged craft.

Kieth Teare translates Anthropic this way: “My product is so bad for mankind that I plan to place an indelible fingerprint that I created it so that everybody can know I was there.”

I don’t subscribe to Ben Thompson’s Stratechery, so I can’t find a passage to quote, so here’s Keith:

Watermarking is Nuts

Ben Thompson’s Stratechery piece belongs at the top of this week’s issue because he nails this truth.

A watermark starts from suspicion. It assumes AI involvement is the important fact, and that the task of institutions is to detect it after the fact. Thompson’s objection is simple and right: a mark may only show that Claude proofread, translated, summarized, or converted human-origin work. A missing mark does not prove AI was absent. It can accuse legitimate human work and miss machine-written work at the same time. I read a piece by my good friend Saul Klein this week and Substack’s partner said it was 100% AI written. It isn’t. Sure Saul probably used AI, but his ideas are solid and clear in the piece. They are not those of the tool he used.

More importantly, asking if AI has been used is the wrong moral question. We do not mark work because a calculator helped with the arithmetic, a compiler helped with the code, a camera helped with the image, or a search engine helped with the facts. We judge the result and the responsibility behind it. Is it true? Is it useful? Is it accountable? Is the author using the tool honestly? The bad acts are fraud, fake sources, fake evidence, fake authorship, hidden manipulation, and unaccountable slop. The bad act is not using AI. The vast majority of AI is used in these “good” ways, not in the “bad” ones.

I hope OpenAI, Google, and the other operators on the great AI frontier see this as a perfect opportunity to stand out in the marketplace by not watermarking text.

Are you willing to admit your inbox has become unmanageable? Here’s how I turned mine into a single daily digest of newsletters. And you can too. Here’s what it looks like:

(Yes, I could turn this into a startup, meco.app is already doing that, but let’s build our own!)

Overview

What you’ll need

• separate Gmail account for newsletters (e.g. name.reads@gmail.com)

• Claude (or other LLM) to write or edit the scripts

• script.google.com (free) to hold/run the two scripts:

  • input.gs

  • output.gs

• firebase.com database (free) to store/serve the ingested emails

• optional: Anthropic API account (paid, by usage) to include a summary at the top

How it works

1. Newsletter arrives in new Gmail → filter applies category label

2. Input script runs every 30 min → saves posts to Firebase with category

3. Output script runs daily → reads Firebase → calls Claude to write summary → sends digest email

Sound good?

Simple Tutorial

Start a new Claude session and tell it:

I want to set up a personal newsletter digest called 4D Daily Digest. I have two Google Apps Script files (input.gs and output.gs) that I’ll paste in when you ask for them.

The system works like this: input.gs runs every 30 minutes as a Google Apps Script in a dedicated Gmail account, watches the inbox for newsletters, and saves them to a Firebase Realtime Database. output.gs runs once a day, reads those posts from Firebase, (optionally) calls the Claude API to write a short summary of what’s trending across all my newsletters, and emails me a digest.

Please help me: 1. Set up a Firebase project and get the database URL 2. Create a Google Apps Script project called Digest with two files, paste in the scripts, and fill in my Firebase URL and Gmail address 3. Set up Gmail labels for my newsletter categories and filters to apply them automatically 4. Authorize and test the scripts 5. Set up triggers to run them automatically

Here are the two scripts to paste when it asks for them:

1. input.gs

2. output.gs

Step-by-Step Tutorial

Want more details?

Step 1: Create a dedicated Gmail account

Create a new Gmail account just for newsletters — this keeps your personal inbox clean and gives the scripts a dedicated inbox to watch.

Emails appear in the “uncategorized” section of the digest until you get around to creating the categorization filters (see step 3).

Step 2: Set up Gmail categories

In your newsletter Gmail account, create labels for each category you want. I like to do them by purpose, for example:

• Get Techy

• Build a Biz

• Get Cultured

• Have Fun

Then, for each newsletter you subscribe to, create a filter that applies the right label:

1. In Gmail’s search, type from:domain.com (e.g. from:platformer.com) and hit enter

2. Click the filters/options button at the right edge of the search bar

3. Click Create filter

4. Check Apply the label and choose the category label

5. Check Skip the Inbox (Archive it) if you want to keep this inbox clear

6. Click Create filter

Bonus: You can quit here, and still have a nice RSS-style browseable inbox:

Want to keep going and get that digest?

Step 3: Set up the database

Firebase is a free database to store and serve the emails.

1. Go to console.firebase.google.com

2. Create a new project, name it Digest or whatever (on the following screens, you can enable AI and analytics if you want, but it’s not required)

3. In the left nav, go to Databases → Realtime Database

   

4. Click Create Database, and then any location, and then start in test mode

5. Your database will open. Click over to the Rules tab, and paste the following so the scripts can access the database (NOTE: this is open access, but for personal use that’s fine):

{
  "rules": {
    ".read": true,
    ".write": true
  }
}

4. Go back to the Data tab. Notice your database URL at the top (it looks like https://your-project-default-rtdb.firebaseio.com), you’ll need that in a minute

Step 4: Set up the Digest scripts

Now we just need to paste things into Google Apps Script, a simple place to write and run JavaScript from the cloud.

1. Log into script.google.com as your newsletter Gmail account

2. Create a new project, call it Digest

Set up the input script

1. You’ll see a default file called Code.gs — rename it to input.gs

2. Paste in the contents of input.gs

3. Update FIREBASE_URL → paste your Firebase database URL

Set up the output script

1. Add a second file: click + next to Files → Script → name it output.gs

2. Paste in the contents of output.gs and update these values at the top:

   • FEEDER_BASE_URL → your Firebase database URL

   • BRIEFING_FROM → your newsletter Gmail address

   • BRIEFING_TO → your personal email address

3. Run the script for the first time to authorize it:

   • In input.gs, select processNewEmails from the function dropdown at the top of the editor

   • Click the Run button (▶)

   • A popup will ask you to review permissions — click Review permissions, choose your newsletter Gmail account, then click Allow

   • Check View → Logs to confirm it ran without errors

Schedule the digest

Now just set up two triggers to run the scripts automatically:

1. Click the clock icon (Triggers) in the left sidebar

2. Click + Add Trigger (bottom right) for the input trigger

   1. function: processNewEmails

   2. Time-driven

   3. Hour timer

   4. Every 30 minutes

   5. Save!

3. Click + Add Trigger for the second trigger:

   1. function: sendDailyBriefings

   2. Time-driven

   3. Day timer

   4. your preferred time

   5. Save!

4. You may be asked to authorize again — click Allow

Step 5 (Optional): Add the summaries

You’ll need an Anthropic API key and some cash in the account. If you skip this, the summaries just won’t appear, the rest of the digest works fine.

1. Go to console.anthropic.com and create an account

2. Add a credit card and purchase some credits (~$5 should last months)

3. Go to API Keys and create a new key

4. Paste it into ANTHROPIC_KEY at the top of output.gs

Step 6: Test it

1. In output.gs, make sure TEST_MODE = true

2. Run sendDailyBriefings manually

   1. Make sure the function dropdown says sendDailyBriefings

   2. Click the Run button

   3. Check your inbox: everything looks good? If not, ask Claude how to revise the scripts to appear how you’d like

Step 7: Ship it!

1. Set TEST_MODE = false and save the script

2. You’ll get the digest tomorrow at your preferred time

Tips

• Where do the links go?

  • To the original article or post on the web. For newsletters that don’t publish a web version, the link goes to the Gmail thread — you’ll need to be logged into your newsletter Gmail account to open those.

• Want to change a newsletter’s category?

  • Go to Settings → See all settings → Filters and Blocked Addresses, find the filter for that sender, click Edit, and change the label it applies.

  • You can also add or rename categories at any time under Settings → See all settings → Labels.

• Want to remove a source?

  • Just unsubscribe from the newsletter, if it’s not coming to your inbox it won’t be in the digest

• Want to pause a source but not unsubscribe?

  • Set active: false on the feed entry in Firebase

• Not seeing a source?

  • Gmail will occasionally flag the digest as spam. Mark it as not spam once and it should stop

Other questions? Let me know!

Immigration and Customs Enforcement (ICE) plans to spend tens of millions of dollars on buying specialized gloves that pulse with electrical energy, which make it easier for agents to subdue people, the Associated Press reported on Tuesday.

A user manual for G.L.O.V.E, or Generated Low Output Voltage Emitter, sold by a company called Compliant Technologies, provides more specifics on how the gloves are supposed to work and allegedly be used.

“No more than two (one pair or equivalent) of G.L.O.V.E. devices should be applied at one time to a single Subject,” the manual reads. 404 Media found multiple copies of the manual online and archived a version here. 

The manual says the gloves may even contribute to “sudden death.” It says “Conductive Distraction and De-escalation Devices” or CD3s may “increase the affects [sic] that can cause sudden death including physiological changes.” It then lists increases in blood pressure, respiration, and heart rates; changes in heart rhythm; and increases in adrenaline.

The manual says the gloves must “come in contact with the subject’s skin” and are not effective through clothing or hair. It also says the gloves “can be degraded on individuals and animals that are extremely hairy.”

Another page includes a diagram of the human body and says officers can “grab anywhere on the body” with the gloves “while avoiding the head, face, throat, chest and groin areas.” If the subject resists, though, another page says to deploy the gloves “as necessary.”

It recommends using the gloves for around 15 seconds, but a demonstration video published by Peacekeeper Products, which also sells the device, shows staff using it for only a second or two with people already yelling in pain.

The manual says users should avoid using the G.L.O.V.E device against elderly people, small children, pregnant women, and “the severely handicapped.”

The Associated Press report was based on a Department of Homeland Security (DHS) procurement record that said the gloves would be issued specifically to Homeland Security Investigations (HSI) and Enforcement and Removal Operations (ERO), the part of ICE focused on deportations. Civil rights advocates expressed alarm at the planned purchase of the technology by an agency that is already widely seen as out of control and using physical force disproportionately.

The CEO of Compliant Technologies Jeff Nikaus published a long Instagram Reel after the assassination of Charlie Kirk. “I believe as a nation we are on a precipice,” he said. He said, “we need supernatural activity” and “we need to repent as a nation,” among other explicit Christian Nationalist ideas.

Compliant Technologies did not immediately respond to a request for comment.