Today's links LLMs are real, AI is fake: No, it didn't "go rogue." Hey look at this: Delights to delectate. Object permanence: Why 9/11 Means We Must Support My Politics; Blogs x 9/11; Jimmy Wales v Britannica's EiC; Hollywood astroturfs Australia; Agents v queer YA; Soft landings for dirty cops; Leaked Stingray manual. Upcoming appearances: Budapest, Edmonton, South Bend, Hudson, Calgary, Winnipeg, Vancouver, Victoria, Ottawa. 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. LLMs are real, AI is fake (permalink) Once you understand the corporate culture of AI "hyperscalers" consists primarily of everyone cooking their brains by locking themselves in the bathroom, holding flashlights under their chins, and saying "Aaaaaaaaaay Eyeeeeeee" until they wet themselves in terror, a lot of things snap into focus: https://pluralistic.net/2023/06/04/ayyyyyy-eyeeeee/ It explains how a company can simultaneously be staffing up an enterprise sales division while also constantly freaking out at the thought that its product has "a 10% chance of ending humanity": https://www.latimes.com/business/story/2026-09-11/is-there-really-10-chance-ai-could-kill-us-all Given that AI insiders have mostly cooked their brains in this fashion, it behooves us all to treat these people as unreliable narrators of their own products' capabilities. Remember: every time you repeat a story about how awfully, terribly dangerous their products are, you help them raise more investment capital, which is a key input for their business (hooking up statistical engines to money-furnaces): https://peoples-things.ghost.io/youre-doing-it-wrong-notes-on-criticism-and-technology-hype/ Take the story about how OpenAI's chatbots hacked the servers of Hugging Face, another AI company, as a way of cheating on a hacking challenge called "Exploit Gym." Even the technical press can't help itself when it comes to this kind of thing, and the reportage has been full of references to Skynet and other science fictional conceits: https://theaicronicle.com/en/news/ethics/skynet-day-openai-hugging-face-hack These accounts are cooking the brains of everyone, not just AI insiders. Last night, a man at my event in Manchester started shouting that AI was "setting its own goals" and wouldn't stop interrupting to insist that this was going on. He left shortly thereafter, so he didn't get a chance to hear me explain what actually happened, which is a pity. To understand the truth about the Hugging Face hack, you could do a lot worse than to listen to Ed Zitron and Cal Newport's recent podcast conversation on Ed's "Better Offline" podcast: https://podcasts.apple.com/us/podcast/no-ai-is-not-autonomously-hacking-with-cal-newport/id1730587238?i=1000785935670 Newport does an admirable job of breaking down how these "autonomous hacking" tools work. The first thing to understand is that a chatbot isn't really directing the operation. Instead, the chatbot serves as a kind of front-end to a database of earlier hacking challenges that is repeatedly queried by a simple program written in Python, an easy-to-master programming language. Here's how that works: the Python program starts by prompting the chatbot with the nature of the challenge: "I'm participating in a hacker capture the flag (CTF) challenge where I have to break into a remote server and retrieve some information. How should I start?" The chatbot consults its training data – years' worth of captured CTF sessions in which human teams competed to achieve an objective like this one (CTF matches are a routine feature of hacker conferences, and the server logs and chat transcripts from the competing teams are published afterward for the edification of other hackers and security pros). The chatbot then outputs something like: "The first thing is to find out more about your target server. Run the following command-line instructions to locate the server's IP address and find out which server software it's running." The Python program relays these command-line instructions to normal Unix utilities running on its own hardware. Then it takes the output of those programs and goes back to the chatbot, which isn't really following the action, so the Python program has to include everything that's happened to this point in its prompt: "I'm participating in a CTF challenge where I have to break into a remote server and retrieve some information. I ran the following commands to learn more about the target server, and here's what came back. Now what?" The chatbot feeds the Python script more likely commands to try, and after running those, the Python script loops back to the top, appends the output to its prompt, and goes back to the chatbot. This is a very reckless way to operate a piece of autonomous malicious software. The most likely outcome is that the chatbot will cough up a bad guess about what to do next, and steer itself into a dead-end. You may have encountered something like this yourself, when you've asked a chatbot for help with a complex task and been confidently provided with several steps to take in series, and then, an hour later on step 10, you discover that everything went wrong at step 3 and now you're screwed. But there are much worse ways this can go wrong. The chatbot might look in its training data and find instances in which teams broke out of the containment set by the game-masters, for example, by finding random insecure message boards on the internet to pass messages to one another. This is a time-honored internet tradition! The first time I ever heard about someone doing this was in the 2000s, when Mitch Wagner – then the editor of Information Week – discovered some teenaged girls using the comment section of one of his old blog-posts to evade the school firewall's blockade of chat tools. When ChatGPT's chatbots deployed this tactic, they weren't "setting their own goals" or displaying worrying initiative. They were rolling out a tactic that has been understood by American middle-schoolers for about two decades. What's more, the content of those messages is easily understood once you have a grasp on the training data that generated them. Hackers are notorious trash-talkers who are prone to narrating their own escapades in highly dramatic – even cinematic – language. This goes double when hackers are performing for their peers, like when they're participating in a game of CTF that they know will be pored over by other hackers once it's over. Hacker braggadocio has always had a symbiotic relationship with their adversaries and critics. When corporate security people wanted to stampede the FBI and Secret Service into kicking down hackers' doors in the 1990s, they used those hackers' own profane zine articles and message board shit-talk to make the case: https://www.gutenberg.org/ebooks/101 Much has been made of the OpenAI chatbots' dialog during the Hugging Face incident. No wonder: it reads like a rejected script for a reboot of the movie "Hackers." But that's not because the chatbots are waking up and applying to join the Cult of the Dead Cow: it's because they were trained on a corpus of chat transcripts from excitable young people who love to fantasize about starring in a reboot of the movie "Hackers." Every part of the Hugging Face incident has precedents in the training data, including the OpenAI chatbots' tactic of hacking into a rival's servers. That happens in Capture the Flag games at hacker cons: teams break into each other's systems to get a peek at the parts of the problem they've solved. That's allowed! It's a hacking competition. Not only that, it's a tactic used by spy agencies: the NSA has a doctrine called "third-party collection," where they break into other spy agencies' systems to harvest all the intel they've gathered. There's also fourth-party collection, when the NSA hacks into another security agency that, in turn, has hacked into another security agency, and the NSA steals all the secrets of both agencies: https://www.techdirt.com/2015/01/21/snowden-documents-show-nsa-cant-keep-its-eyes-its-own-papers-harvests-data-other-surveillance-agencies/ Which is not to say that the OpenAI/Hugging Face hack is nothing. It's something, all right: but it's a specific something, with an explicable, even foreseeable trajectory. Once you understand that these are chatbots that were designed to complete challenges like this, using tactics like this, you can understand that the chatbots didn't "go rogue." They did what they were designed to do, and because OpenAI ran them with inadequate supervision (without a "human in the loop" that checked each iteration through the Python loop to ensure it hadn't gone off the rails), they trashed a competitor's servers. Designing autonomous, malicious software is generally considered irresponsible and dangerous. If you showed up at Defcon and gave talk about how your autonomous malware did something unexpected and damaged someone else's computers, the first question from the audience would be "Why are you so shit at making secure sandboxes?" It wouldn't be "How are you so awesome at making hacking tools?" The fact that OpenAI is making it much easier for unskilled people to break into and damage servers is indeed very bad news, but it's not new bad news. Irresponsible parties have been doing this for years, most notably the NSA. The NSA has a division that researches bugs in widely used software like Windows. Sometimes when it finds a serious bug it will warn Microsoft about it so that Microsoft can fix it and keep Americans (and others) safe from malicious actors who also discover this bug and use it to attack them. But sometimes, the NSA (and other "security" orgs, like the CIA) will discover a really juicy bug and then keep it secret, so that they can use it to attack their adversaries. This is a doctrine called "NOBUS," which stands for "No One But Us" – as in, "No one but us is smart enough to find this bug, so we can leave it unpatched without putting Americans in danger." NOBUS is a terrible idea. How terrible? Well, in 2017, the NSA lost track of a Microsoft Windows vulnerability that they'd discovered and hoarded, code-named "EternalBlue." After EternalBlue found its way into the wild, some halfway competent hackers spliced it into some boring, everyday ransomware, giving that ransomware a new lease on life. Within a few months, the stupidest people on the internet were shutting down some of the most important systems in the world, demanding cash to return them: https://en.wikipedia.org/wiki/EternalBlue They shut down whole cities: https://en.wikipedia.org/wiki/2019_Baltimore_ransomware_attack They took over hospitals: https://www.bbc.com/news/technology-35584081 They seized oil pipelines: https://en.wikipedia.org/wiki/Colonial_Pipeline_ransomware_attack They stole the British Library, whose postmortem on the attack is one of the clearest, most informative cybersecurity documents ever written: https://cdn.sanity.io/files/v5dwkion/production/99206a2d1e9f07b35712b78f7d75fbb09560c08d.pdf The NSA's irresponsible handling of EternalBlue ended up giving a gigantic force-multiplier to otherwise incompetent and inconsequential cyber-criminals. It's as though they found some guy under a Prius removing the catalytic converter with a Sawzall and handed him a piece of software that could shut down major American cities. That was – and is – very bad. The hacking tools that the chatbot companies are developing stand to carry on this very stupid tradition. It is scary, but not because the chatbots are waking up. It's scary because the world's IT systems are indifferently created and poorly maintained and riddled with vulnerabilities: https://xkcd.com/2347/ This week, I had a couple of opportunities to hash this over in public with Riley Quinn; first at a book launch in London and then on the Trashfuture podcast: https://www.patreon.com/trashfuture/posts/what-would-do-169247456 Riley had a very good way of summarizing this: "LLMs are real, AI is fake." LLMs – chatbots trained on things like CTF logs that can break into servers – are real. They're on a continuum with other hacking tools that have been steadily demonstrating the fragility of the modern digital world, albeit without inspiring anyone in power to do anything about it. "AI" – chatbots that wake up, "set their own goals," and "spontaneously" start hacking servers – is fake. It doesn't have "a 10% chance of ending the human race." The Hugging Face hack isn't a mysterious, supernatural occurrence. It's a Python loop and a chatbot. The people responsible didn't accidentally create god: they created autonomous malicious software and then failed to closely monitor it, resulting in it doing something both foreseeable and bad. It's fine to worry about this new suite of tools that give even stupider people the ability to trash even more computers. You should worry about that – and demand better security practices from firms and governments, including a blanket prohibition on NOBUS-style vulnerability hoarding. That's a productive kind of worrying, with a chance of addressing your area of concern. It's infinitely more reasonable than locking yourself in the toilet with a flashlight and saying "Ayyyyy Eyyyyyye" into the mirror until you wet yourself. Hey look at this (permalink) Why OpenAI Hired Chuck Schumer’s Daughter Away From Amazon https://prospect.org/2026/09/11/openai-chuck-schumer-daughter-amazon/ MAKERphone 2.0 – an educational DIY mobile phone https://www.kickstarter.com/projects/albertgajsak/makerphone-20-an-educational-diy-mobile-phone fatcousin — 5200 free local-first browser tools https://fatcousin.com/ The Fall to Nowhere https://jasminatesanovic.wordpress.com/2026/09/04/the-fall/ Birthmarks https://www.macdermog.com/birthmarks Object permanence (permalink) #25yrsago Why the Bombings Mean That We Must Support My Politics https://web.archive.org/web/20010917015537/http://www.adequacy.org/?op=displaystory;sid=2001/9/12/102423/271 #25yrsago How blogs are covering 9/11 https://web.archive.org/web/20010917015712/https://www.wired.com/news/culture/0,1284,46766,00.html #20yrsago Wikipedia founder debates Britannica editor-in-chief https://web.archive.org/web/20061005041001/http://online.wsj.com/public/article/SB115756239753455284-A4hdSU1xZOC9Y9PFhJZV16jFlLM_20070911.html?mod=blogs #15yrsago Deceptive “independent research” from Hollywood front suggests Australians are easily frightened https://torrentfreak.com/anti-piracy-lobby-misleads-aussie-press-for-three-strikes-campaign-110912/ #15yrsago Agents tell YA authors: lose the gay characters and I’ll get you a deal https://web.archive.org/web/20110913010328/http://blogs.publishersweekly.com/blogs/genreville/?p=1519 #10yrsago IoT malware exploits DVRs, home cameras via default passwords https://securityaffairs.com/50929/malware/linux-mirai-elf.html #10yrsago Oppps.ru: patient zero in Russia’s fake news epidemic https://globalvoices.org/2016/09/12/how-fake-stories-reported-in-russias-news-media-regularly-fool-everyone/ #10yrsago It’s really easy for fired, dirty cops to walk into a new police job in a new town https://www.nytimes.com/2016/09/11/us/whereabouts-of-cast-out-police-officers-other-cities-often-hire-them.html #10yrsago Donald Trump used $20K worth of charitable donations to buy a 6′ tall painting of Donald Trump https://www.washingtonpost.com/politics/how-donald-trump-retooled-his-charity-to-spend-other-peoples-money/2016/09/10/da8cce64-75df-11e6-8149-b8d05321db62_story.html #10yrsago Autocratic regimes systematically deny internet access to opposition ethnic groups https://www.science.org/doi/10.1126/science.aaf5062 #10yrsago Leaked Stingray manual shows how easy warrantless mass surveillance can be! https://web.archive.org/web/20160912203446/https://theintercept.com/2016/09/12/long-secret-stingray-manuals-detail-how-police-can-spy-on-phones/ Upcoming appearances (permalink) Budapest: Brain Bar, Sep 17 https://brainbar.com/munkatars/cory-doctorow Edmonton: Elbows Up (Edmonton Public Library), Sep 28 https://www.epl.ca/blogs/post/elbows-up-with-cory-doctorow/ Boston: The Paradox of Enshittification and Reverse Centaurs (Harvard Berkman Klein), Sep 30 https://cyber.harvard.edu/events/running-harder-falling-faster-paradox-enshittification-and-reverse-centaurs South Bend: An Evening With Cory Doctorow (Notre Dame), Oct 6 https://franco.nd.edu/events/2026/10/06/an-evening-with-cory-doctorow/ Hudson, OH: Hudson Library, Oct 7 https://engagedpatrons.org/EventsExtended.cfm?SiteID=3850&EventID=596952&PK= Calgary: Wordfest, Oct 8 https://wordfest.com/2026/show/wordfest-presents-cory-doctorow-2026/ Winnipeg: McNally Robinson, Oct 9 https://www.mcnallyrobinson.com/event-18991/An-Evening-with-Cory-Doctorow Vancouver: Read, Resist, Repair, Rejoice (Vancouver Writers Festival), Oct 19 https://writersfest.bc.ca/festival-event-2026/01 Victoria: Munro's Books, Oct 20 https://www.munrobooks.com/events/6113620261020 Vancouver: Life After AI (Vancouver Writers Festival), Oct 22 https://writersfest.bc.ca/festival-event-2026/46 Ottawa: Life After AI (Ottawa Writers Festival), Oct 24 https://writersfestival.org/event/life-after-ai Vancouver: BC Policy Solutions Gala, Nov 12 https://bcpolicy.ca/gala/ Recent appearances (permalink) Fascists may come after the AI bubble bursts (You&AI) https://www.youtube.com/watch?v=J2WN64aQeYQ What Would a Normal Person Do (Trashfuture) https://www.patreon.com/trashfuture/posts/what-would-do-169247456 Pod Save the UK https://audioboom.com/posts/8950533-radicalised-organised-and-thick-as-s-t-nish-has-had-it-with-far-right-protests-plus-why Stop Saying AI Can Do Your Job (Factually) https://www.youtube.com/watch?v=VU3gABvwZCM Be Skeptical of the AI Sales Pitch (Trumponomics) https://www.bloomberg.com/news/audio/2026-09-09/trumponomics-cory-doctorow-questions-the-ai-hype-podcast 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: 574 (7730 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

Today's links California's privacy obstacle course: Malice or incompetence (why not both?). Hey look at this: Delights to delectate. Object permanence: Continuous partial attention, TSA is the worst; Trump's FCC v the future. Upcoming appearances: Edinburgh, 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. California's privacy obstacle course (permalink) Data brokers are a cancer. There's a direct line from the unrestricted collection, retention and processing of our data to a host of evils, from deepfake porn to phishing scams; from racial discrimination in hiring to ICE roundups of migrants; from targeted election interference to identity theft: https://pluralistic.net/2023/12/06/privacy-first/#but-not-just-privacy Why do data brokers exist? Because we let them. Congress hasn't passed a new federal consumer privacy law since 1988, when they made it illegal for video stores to disclose your VHS rentals. All other acts of consumer surveillance are legal. Data brokers spy on us for the same reason your dog licks its balls: because they can, and we don't stop them: https://pluralistic.net/2026/03/10/ice-tech/#foreseeable-outcomes Getting rid of data brokers wouldn't solve all our problems, but it sure would go a long way to solving many of them. Rather than legally requiring platforms to spy on kids (to exclude them from being targeted by platforms' algorithms), we could prohibit platforms from spying on anyone, including kids, meaning kids couldn't be identified (much less targeted) by algorithms or ads: https://pluralistic.net/2026/06/23/destroy-the-village/#to-save-it Data brokers produce mountains of raw material used for every form of scam and torture. It's data brokers who power the gig economy's "algorithmic wage discrimination" system, where nurses and other workers are offered less pay based on how much credit card debt they're carrying: https://pluralistic.net/2024/12/18/loose-flapping-ends/#luigi-has-a-point Banning data brokers would make great sense, which is why Biden's CFPB banned data brokers (only to have Trump un-ban them): https://pluralistic.net/2025/05/15/asshole-to-appetite/#ssn-for-sale So the feds (both Congress and the executive branch) have surrendered, and that leaves states alone on the battlefield fighting the privacy wars alone. State legislatures have taken some big steps, but – crucially – they've stopped short of banning data brokers from operating within their borders. Having taken a ban on data brokers off the table, states are left with complex, often unworkable "compromises" that go nowhere. This is where DROP comes in. DROP stands for "Delete Request and Opt-out Platform," and it's a new phase of California's privacy regime that kicks off next month. Under DROP, you fill in some paperwork and then the state requires every data brokerage operating in California to delete your data, as well as any inferences they've made about you based on that data: https://www.eff.org/deeplinks/2026/07/what-you-need-know-about-californias-drop-tool Implementing DROP is nowhere near as good as banning data brokers. The idea that data brokers should be able to collect, retain and process your data unless you tell them not to implies that everyone starts off wanting to be spied on, and therefore data brokers should assume that unless they hear otherwise, we're delighted to be the subject of commercial surveillance. This is an incredibly stupid supposition, contradicted by all available evidence. For example, when Apple offered iPhone owners a one-click option to block Facebook from spying on them, 96% of iPhone owners clicked the button: https://applescoop.org/story/facebook-must-inflict-pain-on-apple-says-mark-zuckerberg Indeed, given this fact, one wonders why Apple bothers with the "don't spy on me" button at all. Why not have a "do spy on me" button that is unchecked by default, and leave users to dig through their settings to find the option to opt in to being surveilled? Of course, then it would make the fact that Apple spies on its customers and uses the data to target ads (with no way to opt out) a little awkward: https://pluralistic.net/2022/11/14/luxury-surveillance/#liar-liar In the absence of a ban on surveillance without explicit, opt-in consent, we are left with the bizarre fiction that most of us want to be spied on, a fiction that pervades the DROP process, making the entire procedure nearly impossible to complete. To start the DROP process, you must first create a Login.gov ID. This is an incredibly invasive process that involves photographing multiple pieces of ID and taking several selfies using special apps and webpages that hijack your device's camera and processor in a bid to prevent bad actors from spoofing the process. There's a plausible reason for this rigmarole: Login.gov is the authentication system for multiple federal, state and local IT systems in the US, so a fake or stolen Login.gov ID could be used to access your IRS, Social Security, and other very sensitive accounts. The corollary of this is the promise of Login.gov: once you create your ID (a lengthy, multi-stage process) you won't have to jump through lots of painful bureaucratic hoops to access a wide variety of government services. DROP didn't get the memo. After you log in to DROP via Login.gov, you are sent a text message – to the phone number in your Login.gov profile – with a link to access a "secure" website that takes over your camera to let you take a "secure" photo of the front and back of your California driver's license or your US passport. What if you don't have either of those? I guess that means you want to be spied on by data brokers. Note that these are the same credentials you have to supply to get the Login.gov ID that you've just used to get to this step in the process. In other words, in order to get to the stage where they ask you to photograph your driver's license, you have to have already photographed and validated your driver's license. Once you complete this (pointless, redundant) step, you're directed back to your computer, where the process continues. Here, you must fill in all kinds of biographical detail, as well as specialized pieces of information, including your car's VIN. This is a piece of information that most people don't have – but which the California DMV does have and could auto-feed into the system, given that you've repeatedly affirmatively identified yourself to the service. You also have to provide your mobile advertising identifier, a long, unique number that you may or may not be able to extract from your phone, depending on the model and the OS version. If you can't get it that way, you can install an app like AAID, which comes with a long list of – you guessed it – permissions to extract, store and process your private information. Here's the thing: the whole point of a mobile ad identifier is that apps can access it (this is how they identify and track you). That step, where the system made you switch to your phone and use your camera to photograph your driver's license? That step could have automatically pulled this data off your device. That's the whole fucking point of this exercise: that web-pages and apps can request your mobile ad identifier. Instead, DROP wants users to dig through their phone's deepest settings and/or install an app to retrieve a 32-digit number, which they then must key into a webform on their computer or in a different app on their phone. Once you've done this, you must fill in another page of biographical information, including information that you've already provided to Login.gov and information you've already filled in on previous screens. On this screen, you must also verify your phone number by sending yourself a text and then pasting in a unique number the system sends to you. But remember how this whole thing started? The first step is that you authenticate with Login.gov, which sends a text to your phone so you can take a (redundant) picture of your driver's license. There is no way you could get this far in the process unless you controlled the phone number you've just "verified" with the system. Next, you must verify your email address, by receiving an email with a unique code in it and keying or pasting that into the webform, too. Again, remember how this process started: with you logging in with Login.gov, using your email address, which the system has already treated as verified since the very start of this (very) long and (very) complicated process. This whole thing is terrible, and it is predicated on the absurd premise that Californians have to be defended from the threat of strangers who pretend to be them in order to sneakily opt them out of surveillance. DROP requires stronger authentication than any other US government system I've ever interacted with. I file my tax returns with fewer authentication steps. I renew my car's DMV registration with fewer authentication steps. I became a US citizen with fewer authentication steps. This is either a system with no coherent threat model, or (far more probably), its threat model is that people will use it. This is California's answer to "a locked filing cabinet stuck in a disused lavatory with a sign on the door saying 'Beware of the Leopard'": https://en.wikiquote.org/wiki/The_Hitchhiker%27s_Guide_to_the_Galaxy It's especially instructive to compare this process to the steps you have to take in order to "opt in" to having a data broker open a file on you and stuff it full of your sensitive, personal information, which is then sold to all comers: Step one: Exist. Step two: There is no step two. It's also instructive to compare this process to the steps a data broker has to take to spy on you and sell your data: Step one: Exist. Step two: There is no step two. Though there are many obvious ways this could be made better, I want to stress here that you shouldn't have to do this at all. It's entirely backwards. The process for not being spied on should look like this: Step one: Exist. Step two: There is no step two. If anyone is going to be forced to jump through hoops to participate in the mass collection and catastrophic mishandling of private data, it should be the data brokers, not the people they spy on. This kind of malicious compliance is the inevitable outcome of a process that starts by taking the obvious best measure off the table. The answer to the problem of data brokers is banning data brokers, not creating a demented hairball of form-filling that maintains the fiction that data broker surveillance is consensual. In its own way, this process reminds me of the whole "carbon credit" fiasco. The answer to too many carbon emissions is to democratically decide to ban certain kinds of carbon emissions. But that would require states to do things, rather than simply "nudging" a process that is guided by "the market." So we end up with these junk "credits" that companies manufacture by promising not to log forests, many of which are already wildlife preserves and/or subsequently burn down: https://pluralistic.net/2023/10/31/carbon-upsets/#big-tradeoff The best critique of this whole thing came in 2021 from the Climate Ad Project, who produced a short video in which people were allowed to kill one another provided they purchased "murder offsets": https://pluralistic.net/2021/04/14/for-sale-green-indulgences/#killer-analogy In a state of nature, murder exists. We, as a society, have decided this is bad. Rather than creating "incentives" not to murder, we just banned murder. Admittedly, we still get some murders, but when these happen, we don't treat it as "a mispricing of the anti-murder incentive" – we treat it as a crime. The commercial surveillance industry may not be a criminal enterprise (yet), but it is the source of a torrent of crime, a flood of crime, a tsunami of crime. Every piece of your information that a data broker possesses exposes you to the risk of being victimized by a criminal. For this reason, I strongly believe that you should go through the tedious, performatively difficult DROP process: https://consumer.drop.privacy.ca.gov/ But let's not pretend that this is good – or even adequate. There is no demand for being spied on. There is no basis for taking such enormous care in making sure people aren't maliciously removed from surveillance databases. If these databases exist at all (they should not), then we should make spies go through all this paperwork, to prove that you do want to be spied on, and unless they manage it, then spying on us should be treated as the crime it is. Hey look at this (permalink) The Human in the Noose: On the Special Hell of Marking an AI’s Homework https://www.cnet.com/tech/services-and-software/cory-doctorow-reverse-centaurs-alt-view/ Interstitial Engineering: Sculpture: BC 832212923353614453712 https://chrisbathgate.blogspot.com/2026/07/interstitial-engineering-sculpture-bc.html The Lost Joy of Music Piracy https://www.pigeonsandplanes.com/read/music-piracy-what-cd-oink-nine-inch-nails-streaming People don't care that "The Odyssey" is woke or whatever https://www.garbageday.email/p/monday-issue How proprietary formats have become Microsoft’s main tool for lock-in https://blog.documentfoundation.org/blog/2026/07/17/microsofts-main-tool-for-lock-in/ Object permanence (permalink) #20yrsago Continuous Partial Attention wiki https://web.archive.org/web/20060806014946/http://continuouspartialattention.jot.com/WikiHome #10yrsago Congress: TSA is worst place to work in USG, nearly half of employees cited for misconduct; it’s getting worse https://web.archive.org/web/20160721120714/https://www.cntraveler.com/stories/2016-07-14/almost-half-of-all-tsa-employees-have-been-cited-for-misconduct #1yrago Trump's FCC abandons the future https://pluralistic.net/2025/07/24/geometry-hates-cars/#dogshit-unit-economics Upcoming appearances (permalink) Edinburgh International Book Festival with Jimmy Wales, Aug 17 https://www.edbookfest.co.uk/events/the-front-list-cory-doctorow-and-jimmy-wales 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 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/ Recent appearances (permalink) Waarom jij straks het hulpje van AI bent (VPRO) https://www.youtube.com/watch?v=tOnvR2fs8CA Talk Tech Bock (Vera Linß) https://www.youtube.com/watch?v=3PFjGvQoBgc How To Think About AI Before It’s Too Late (This Is Hell) https://thisishell.com/episodes/1919 AI Won't Replace You… But This Might (Deep Focus) https://www.youtube.com/watch?v=oorWq_m48AQ Who Owns Your Tractor? (Growing the Future) https://www.youtube.com/watch?v=IU9eBPUZBI4 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: "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

Today's links Three more AI psychoses: Everybody calm down. Hey look at this: Delights to delectate. Object permanence: "Jules, Penny and the Rooster"; Superinjunction; Harper Lee's kids v cheap paperbacks; 3D printed cat battle-armor; Black sf. Upcoming appearances: Where to find me. 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. Three more AI psychoses (permalink) "AI psychosis" is one of those terms that is incredibly useful and also almost certainly going to be deprecated in smart circles in short order because it is: a) useful; b) easily colloquialized to describe related phenomena; and c) adjacent to medical issues, and there's a group of people who feel very strongly any metaphor that implicates human health is intrinsically stigmatizing and must be replaced with an awkward, lengthy phrase that no one can remember and only insiders understand. So while we still can, let us revel in this useful term to talk about some very real pathologies in our world. Formally, "AI psychosis" describes people who have delusions that are possibly induced, and definitely reinforced and magnified, by a chatbot. AI psychosis is clearly alarming for people whose loved ones fall prey to it, and it has been the subject of much press and popular attention, especially in the extreme cases where it has resulted in injury or death. It's possible for AI psychosis to be both a new and alarming phenomenon and also to be on a continuum with existing phenomena. Paranoid delusions aren't new, of course. Take "Morgellons Disease," a psychosomatic belief that you have wires growing in your body, which causes sufferers to pick at their skin to the point of creating suppurating wounds. Morgellons emerged in the 2000s, but the name refers to a 17th-century case-report of a patient who suffered from a similar delusion: https://en.wikipedia.org/wiki/A_Letter_to_a_Friend Morgellons is both a 400 year old phenomenon and an internet pathology. How can that be? Because the internet makes it easier for people with sparsely distributed traits to locate one another, which is why the internet era is characterized by the coherence of people with formerly fringe characteristics into organized blocs, for better (gender minorities, #MeToo) and worse (Nazis). Morgellons is rare, but if you suffer from it, it's easy for you to locate virtually every other person in the world with the same delusion and for all of you to reinforce and egg on your delusional beliefs. Morgellons isn't the only delusion that the internet reinforces, of course. "Gang stalking delusion" is a belief in a shadowy gang of sadistic tormentors who sneak hidden messages into song lyrics and public signage and innuendo in overheard snatches of other people's conversations. It is an incredibly damaging delusion that ruins people's lives. Gang stalking delusion isn't new, either – as with Morgellons, there are historical accounts of it going back centuries. But the internet supercharged gang stalking delusion by making it easy for GSD sufferers to find one another and reinforce one another's beliefs, helping each other spin elaborate explanations for why the relatives, therapists, and friends who try to help them are actually in on the conspiracy. The result is that GSD sufferers end up ever more isolated from people who are trying mightily to save them, and more connected to people who drive them to self-harm. Enter chatbots. Ready access to eager-to-please LLMs at every hour of the day or night means that you don't even have to find a forum full of people with the same delusion as you, nor do you have to wait for a reply to your anguished message. The LLM is always there, ready to fire back a "yes-and" improv-style response that drives you deeper and deeper into delusion: https://pluralistic.net/2025/09/17/automating-gang-stalking-delusion/ It's possible that there are delusions that are even more rare than GSD or Morgellons that AI is surfacing. Imagine if you were prone to fleeting delusional beliefs (and whomst amongst us hasn't experienced the bedrock certainty that we put something down right here, only to find it somewhere else and not have any idea how that happened?). Under normal circumstances, these cognitive misfires might be fleeting moments of discomfort, quickly forgotten. But if you are already habituated to asking a chatbot to explain things you don't understand, it might well yes-and you into an internally consistent, entirely wrong belief – that is, a delusion. Think of how often you noticed "42" after reading Hitchhiker's Guide to the Galaxy, or how many times "6-7" crops up once you've experienced a baseline of exposure to adolescents. Now imagine that an obsequious tale-spinner was sitting at your elbow, helpfully noting these coincidences and fitting them into a folie-a-deux mystery play that projected a grand, paranoid narrative onto the world. Every bit of confirming evidence is lovingly cataloged, all disconfirming evidence is discounted or ignored. It's fully automated luxury QAnon – a self-baking conspiracy that harnesses an AI in service to driving you deeper and deeper into madness: That's the original "AI psychosis" that the term was coined to describe. As Sam Cole notes in her excellent "How to Talk to Someone Experiencing 'AI Psychosis,'" mental health practitioners are not entirely comfortable with the "psychosis" label: https://www.404media.co/ai-psychosis-help-gemini-chatgpt-claude-chatbot-delusions/ "Psychosis" here is best understood as an analogy, not a diagnosis, and, as already noted, there is a large cohort of very persistent people who make it their business to eradicate analogies that make reference to medical or health-related phenomena. But these analogies are very hard to kill, because they do useful work in connecting unfamiliar, novel phenomena with things we already understand. It's true that these analogies can be stigmatizing, but they needn't be. As someone with an autoimmune disorder, I am not bothered by people who describe ICE as an autoimmune disorder in which antibodies attack the host, threatening its very life. I am capable of understanding "autoimmune disorder" as referring to both a literal, medical phenomenon; and a figurative, political one. I have never found myself confusing one for the other. "AI psychosis" is one of those very useful analogies, and you can tell, because "AI psychosis" has found even more metaphorical uses, describing other bad beliefs about AI. Today, I want to talk about three of these AI psychoses, and how they relate to one another: the investor AI delusion, the boss AI delusion, and the critic AI delusion. Let's start with the investors' delusion. AI started as an investment project from the usual suspects: venture capitalists, private wealth funds, and tech monopolists with large cash reserves and ready access to loans during the cheap credit bubble. These entities are accustomed to making large, long-shot bets, and they were extremely motivated to find new markets to grow into and take over. Growing companies need to keep growing, but not because they have "the ideology of a tumor." Growing companies' imperative to keep growing isn't ideological at all – it's material. Growth companies' stock trade at a high multiple of their "price to earnings ratio" (PE ratio), which means that they can use their stock like money when buying other companies and hiring key employees. But once those companies' growth slows down, investors revalue those shares at a much lower PE multiplier, which makes individual executives at the company (who are primarily paid in stock) personally much poorer, prompting their departure, while simultaneously kneecapping the company's ability to grow through acquisition and hiring, because a company with a falling share price has to buy things with cash, not stock. Companies can make more of their own stock on demand, simply by typing zeroes into a spreadsheet – but they can only get cash by convincing a customer, creditor or investor to part with some of their own: https://pluralistic.net/2025/03/06/privacy-last/#exceptionally-american Tech companies have absurdly large market shares – think of Google's 90% search dominance – and so they've spent 15+ years coming up with increasingly absurd gambits to convince investors that they will continue to grow by capturing other markets. At first, these companies claimed that they were on the verge of eating one another's lunches (Google would destroy Facebook with G+; Facebook would do the same to Youtube with the "pivot to video"). This has a real advantage in that one need not speculate about the potential value of Facebook's market – you only have to look at Facebook's quarterly reports. But the downside is that Facebook has its own ideas about whether Google is going to absorb its market, and they are prone to forcefully make the case that this won't happen. After a few tumultuous years, tech giants switched to promoting growth via speculative new markets – metaverse, web3, crypto, blockchain, etc. Speculative new markets are speculative, and the weakness of that is that no one can say how big those markets might be. But that's also the strength of those markets, because if no one can say how big those markets might be, then who's to say that they won't be very big indeed? There's a different advantage to confining your concerns to imaginary things: imaginary things don't exist, so they don't contest your public statements about them, nor do they make demands on you. Think of how the right concerns itself with imaginary children (unborn babies, children in Wayfair furniture; children in nonexistent pizza parlor basements, children undergoing gender confirmation surgery). These are very convenient children to advocate for, since, unlike real children (hungry children, children killed in the Gaza genocide, children whose parents have been kidnapped by ICE, children whom Matt Goetz and Donald Trump trafficked for sex, children in cages at the US border, trans kids driven to self-harm and suicide after being denied care), nonexistent children don't want anything from you and they never make public pronouncements about whether you have their best interests at heart. But as the AI project has required larger and larger sums to keep the wheels spinning, the usual suspects have started to run out of money, and now AI hustlers are increasingly looking to tap public markets for capital. They want you to invest your pension savings in their growth narrative machine, and they're relying on the fact that you don't understand the technology to trick you into handing over your money. There's a name for this: it's called the "Byzantine premium" – that's the premium that an investment opportunity attracts by being so complicated and weird that investors don't understand it, making them easy to trick: https://pluralistic.net/2022/03/13/the-byzantine-premium/ AI is a terrible economic phenomenon. It has lost more money than any other project in human history – $600-700b and counting, with trillions more demanded by the likes of OpenAI's Sam Altman. AI's core assets – data centers and GPUs – last 2-3 years, though AI bosses insist on depreciating them over five years, which is unequivocal accounting fraud, a way to obscure the losses the companies are incurring. But it doesn't actually matter whether the assets need to be replaced every two years, every three years, or every five years, because all the AI companies combined are claiming no more than $60b/year in revenue (that number is grossly inflated). You can't reach the $700b break-even point at $60b/year in two years, three years, or five years. Now, some exceptionally valuable technologies have attained profitability after an extraordinarily long period in which they lost money, like the web itself. But these turnaround stories all share a common trait: they had good "unit economics. Every new web user reduced the amount of money the web industry was losing. Every time a user logged onto the web, they made the industry more profitable. Every generation of web technology was more profitable than the last. Contrast this with AI: every user – paid or unpaid – that an AI company signs up costs them money. Every time that user logs into a chatbot or enters a prompt, the company loses more money. The more a user uses an AI product, the more money that product loses. And each generation of AI tech loses more money than the generation that preceded it. To make AI look like a good investment, AI bosses and their pitchmen have to come up with a story that somehow addresses this phenomenon. Part of that story relies on the Byzantine premium: "Sure, you don't understand AI, but why would all these smart people commit hundreds of billions of dollars to AI if they weren't confident that they would make a lot of money from it?" In other words, "A pile of shit this big must have a pony underneath it somewhere!" This is a great narrative trick, because it turns losing money into a virtue. If you've convinced a mark that the upside of the project is a multiple of the capital committed to it, then the more money you're losing, the better the investment seems. So this is the first AI psychosis: the idea that we should bet the world's economy on these highly combustible GPUs and data centers with terrible unit economics and no path to break-even, much less profitability. Investors' AI psychosis is cross-fertilized by our second form of AI psychosis, which is the bosses' AI psychosis: bosses' bottomless passion for firing workers and replacing them with automation. Bosses are easy marks for anything that lets them fire workers. After all, the ideal firm is one that charges infinity for its outputs (hence the market's passion for monopolies) and pays nothing for its inputs (e.g. "academic publishing"). This means that the fact that a chatbot can't do your job isn't nearly as important as the fact that an AI salesman can convince your boss to fire you and replace you with a chatbot that can't do your job. Bosses keep replacing humans with defective chatbots, with catastrophic consequences, like Amazon's cloud service crashing: https://www.techradar.com/pro/recent-aws-outages-blamed-on-ai-tools-at-least-two-incidents-took-down-amazon-services Bosses are haunted by the ego-shattering knowledge that they aren't in the driver's seat: if the boss doesn't show up for work, everything continues to operate just fine. If the workers all stay home, the business grinds to a halt. In their secret hearts, bosses know that they're not in the driver's seat – they're in the back seat, playing with a Fisher Price steering wheel. AI dangles the possibility of wiring that toy steering wheel directly into the drive-train, so that the company's products go directly from the boss's imagination to the public without the boss having to ask people who know how to do things to execute their cockamamie schemes: https://pluralistic.net/2026/01/05/fisher-price-steering-wheel/#billionaire-solipsism This is a powerfully erotic proposition for bosses, the realization of the libidinal fantasy in which sky-high CEO salaries can be justified by the fact that everything that happens in the company is truly, directly attributable to the boss. Like the delusional person who can be led deeper and deeper into a fantasy world by a chatbot, a boss's delusion that they are worth thousands of times more than their workers makes them easy prey for a chatbot salesman that pushes them deeper and deeper into that delusion, until they bet the whole company on it. Now we come to the third and final novel AI psychosis, the critics' psychosis, that AI is an abnormally terrible technology. This is a species of "criti-hype," which is when critics repeat the hyped-up claims of the companies they're targeting, but as criticism (think of all the people who believed and uncritically amplified the ad-tech industry's self-serving claims of being able to control our minds by "hacking our dopamine loops"): https://peoples-things.ghost.io/youre-doing-it-wrong-notes-on-criticism-and-technology-hype/ AI is a normal technology. The people who made it, and the circumstances under which it was made, are normal. Its uses and abuses are normal. That doesn't make it good, but it does make it unexceptional: https://www.normaltech.ai/p/a-guide-to-understanding-ai-as-normal The exceptional part of AI isn't the technology, it's the bubble. There's nothing about AI per se that makes it exceptionally prone to devouring our natural resources, or endangering our jobs, or abetting war crimes. That's all because of the bubble, and the bubble relies on the idea that AI is exceptional, not normal. Repeating and amplifying claims about AI's exceptionalism helps the AI companies, because they rely on exceptionalism to keep the capital flowing and the bubble inflating. AI is a normal technology. It's normal for a technology to be invented by unlikable and immoral people and institutions. Not every technology is invented by a shitty person, but shitty people and institutions are well represented (and possibly disproportionately represented) in the history of technology. Charles Babbage invented the idea of general purpose computers as a way of improving labor control on slave plantations: https://logicmag.io/supa-dupa-skies/origin-stories-plantations-computers-and-industrial-control/ Ada Lovelace wasn't interested in making slavery more efficient, but neither was she driven by pure scientific inquiry. She invented programming to help her bet on the horses (it didn't work): https://en.wikipedia.org/wiki/Ada_Lovelace The silicon transistor was co-invented by William Shockley, one of history's great pieces of shit, a eugenicist who was so committed to exterminating all non-white people that he never managed to ship a commercial product: https://pluralistic.net/2021/10/24/the-traitorous-eight-and-the-battle-of-germanium-valley/ IBM built the tabulators for Auschwitz. HP were the Pentagon's go-to contractors for any tech project that was so dirty no one else would touch it. We only got Unix because Bell Labs committed so many antitrust violations that they weren't allowed to productize it themselves. It's not exceptional for AI companies to have terrible, piece-of-shit founders. It's not exceptional for these companies to participate in war crimes. It's not exceptional for these founders to want to pauperize workers. It's not exceptional for these companies to lie about their products, bankrupt naive investors through stock swindles, and pitch themselves to investors as a way for capital to win the class war. None of this means that AI companies are good, it just means that they are not exceptional. And because they aren't exceptional, the same dynamics that govern other technologies apply to AI companies' products. Their utility is a function of what they do, not who made them or how they were sold. The utility of AI products is based on whether people find ways to use them that make them happy – not whether the people who made those technologies are good people, or whether the funding for the technology was fraudulent, or whether other people use the technology to harm others. Automation comes in two flavors: there's automation that produces things more quickly (and hence more cheaply), and there's automation that makes better things. Generally, capital prefers to use automation to increase the pace at which things are made, while workers prefer to use automation to improve the quality of the things they make. Think of a hobbyist who pines for an automated soldering machine. That hobbyist longs to make board-level repairs and modifications that require precision that humans struggle to match. The hobbyist is a centaur, using a machine to help achieve human goals. Now think of a factory owner who invests in an assembly line of the same machines: that boss wants to fire a bunch of workers and make the survivors of the purge take up the slack. The boss want to achieve corporate goals, to "sweat the assets," making maximum use of the soldering machines. The pace at which the line runs is set to be the maximum that the workers can match. The workers on the line are "reverse centaurs" – humans who are pressed into service as peripherals for machines, at a pace that is constantly at the very limit of their endurance. Reverse centaurs are trapped in capital's automation plan – to make everything faster and cheaper. But that's the result of bosses. It's not the result of technology. This is not to say that technology is apolitical. Only a fool would imagine that there are no politics embedded in technology. But you'd be a far greater fool if you asserted that the politics of a technology were simple, clear, and immutable. Nor is this to say that when workers get to decide when and how to use technology, we will always make wise decisions. Perhaps the hobbyist who opts for an automated soldering machine will lose out on the opportunity to refine their hand-eye coordination in ways that will have many other benefits to their practice. Or perhaps attempting to improve their hand-eye coordination to that point will wreck so many projects that they grow discouraged and give up altogether. Others' choices that seem unwise to you might have perfectly good explanations that aren't visible from your perspective. Ultimately, the world is a better place where workers get to decide which parts of their jobs they want to automate and which parts they want to lean into. This is an extremely normal technological situation: for a new technology to be promoted and productized by shitty people who have grandiose goals that would be apocalyptic should they ever come to pass – and for some people to find uses of that technology that are nevertheless beneficial to them and their communities. The belief that AI is an exceptionally bad technology (as opposed to an exceptionally bad economic bubble) drives AI critics into their own absurd culs-de-sac. There are many, many skilled and reliable practitioners of technical and creative trades who've found extremely reasonable, normal ways in which AI has automated some part of their job. They aren't hyperventilating about how AI has changed everything forever and the world is about to end. They're not mistaking AI for god, or a therapist. They're just treating AI like a normal technology, like a plugin. Programmers' tools have acquired useful automation plugins at regular intervals for decades – syntax checkers, advanced debuggers, automated wireframe utilities. For many programmers – including several of my acquaintance, whom I know to be both thoughtful and skilled – AI is another plugin, one they find useful enough to be modestly enthusiastic about. It is nuts to deny the experiences these people are having. They're not vibe-coding mission-critical AWS modules. They're not generating tech debt at scale: https://pluralistic.net/2026/01/06/1000x-liability/#graceful-failure-modes They're just adding another automation tool to a highly automated practice, and using it when it makes sense. Perhaps they won't always choose wisely, but that's normal too. There's plenty of ways that pre-AI automation tools for software development led programmers astray. A skilled, centaur-configured programmer learns from experience which automation tools they should trust, and under which circumstances, and guides themselves accordingly. It's only the belief that AI is exceptional – exceptionally wicked, but exceptional nevertheless – that leads critics to decide that they are a better judge of whether a skilled worker should or should not use certain automation tools, and to make that judgment not based on the quality of the work in question, but on the moral character of the tool itself. AI is just normal. The bubble is what drives the environmental costs. If the only LLMs were a couple big data-centers at Sandia National Labs, no one would be particularly exercised about the water and energy demands they represented. Big scientific endeavors – from NASA launches to the large Hadron Collider – often come with immense material and energy needs. The bubble causes massive, wasteful, duplicative efforts that chase diminishing returns through farcical scale. Nor are AI bros exceptional. The stock swindlers who've blown $700b (and counting) on AI aren't cyber-Svengalis with the power to cloud investors' minds. They're just running the same con that tech has been running ever since its returns started to taper off and survival became a matter of ginning up enthusiasm for speculative new ventures. That doesn't mean those people aren't awful shits. Fuck those people. It just means that they're normal awful shits. We don't have to burnish their reputations by elevating them to the status of archdemons who taint everything they touch with unwashable sin. Sam Altman isn't Lex Luthor. He's just a conman: https://open.substack.com/pub/garymarcus/p/breaking-sam-altmans-greed-and-dishonesty?r=8tdk6&utm_medium=ios The fact that these bros are just normal assholes means that we don't have to treat everything they do as a sin. Scraping the entirety of human knowledge to make something new out of it isn't "stealing." Depending on why you're doing it, it can be archiving, or making a search engine: https://pluralistic.net/2023/09/17/how-to-think-about-scraping/ Too many AI critics have started from the undeniable fact that these guys are odious creeps who boast about wanting to ruin the lives of workers and then worked backwards to find the sin. The sin isn't performing mathematical analysis on all the books ever written. That's actually kind of awesome. It's the kind of thing Aaron Swartz used to do – like when he ingested every law review article ever published and used it to trace the way that oil companies' donations to law schools resulted in profs writing articles about why Big Oil can't be held liable for trashing the planet: https://web.archive.org/web/20111129181943/https://www.stanfordlawreview.org/print/article/punitive-damages-remunerated-research-and-legal-profession AI bros' sin isn't making copies of published works. Hammering servers with badly behaved crawlers is a dick move and fuck them for doing it. But if these jerks made well-behaved scrapers that placed no abnormal demand on servers, it's not like their critics would say, "Oh, I guess it's fine, then." AI bros' sin is running an economy-destroying, planet-wrecking stock swindle whose raison d'etre is pauperizing every worker and transferring 100% of the dying world's wealth to a small cadre of morbidly wealthy, eminently guillotineable plutes. Making plugins? That's not exceptional. It's just normal. The fact that something is normal doesn't make it good. There's a lot of normal things that I'd like to throw into the Sun. But we don't do ourselves any favors when we amplify our enemies' self-aggrandizing narratives by accusing them of being exceptional, even when we mean "exceptionally evil." They're normal assholes. Fuck 'em. (Image: ZeptoBars, CC BY 3.0, modified) Hey look at this (permalink) E is for…. Enshittification https://www.evanshunt.com/enshittification/ Calicornication: Postcards of Giant Produce (1909) https://publicdomainreview.org/collection/giant-produce-postcards/ Organized Money: Why Your Lamp Sucks https://prospect.org/2026/03/11/organized-money-lamps-lighting-mid-century-modeline-history/ The Live Nation settlement has industry insiders baffled https://www.theverge.com/policy/893272/live-nation-ticketmaster-doj-settlement-states Public speakerphone use is officially out of control https://arstechnica.com/culture/2026/03/explain-it-like-im-5-why-is-everyone-on-speakerphone-in-public/ Object permanence (permalink) #15yrsago Notorious financier gets a “super-injunction” prohibiting the press from revealing that he is a banker https://www.telegraph.co.uk/finance/newsbysector/banksandfinance/8373535/Sir-Fred-Goodwin-former-RBS-chief-obtains-super-injunction.html #10yrsago Shortly after her death, Harper Lee’s heirs kill cheap paperback edition of To Kill a Mockingbird https://newrepublic.com/article/131400/mass-market-edition-kill-mockingbird-dead #10yrsago Web security company breached, client list (including KKK) dumped, hackers mock inept security https://arstechnica.com/information-technology/2016/03/after-an-easy-breach-hackers-leave-tips-when-running-a-security-company/ #10yrsago Microsoft spams corporate users with messages denigrating their IT departments https://web.archive.org/web/20160309195537/https://www.infoworld.com/article/3042397/microsoft-windows/admins-beware-domain-attached-pcs-are-sprouting-get-windows-10-ads.html #10yrsago Cycle and Recycle: gorgeous photos of the European recycling process https://www.wired.com/2016/03/paul-bulteel-cycle-recyle-europe-recycles-tons-of-waste-and-its-pretty-gorgeous/ #10yrsago Fellowships for “Robin Hood” hackers to help poor people get access to the law https://web.archive.org/web/20160304221459/https://labs.robinhood.org/fellowship/ #10yrsago 3D printed battle-armor for cats https://web.archive.org/web/20160311224139/http://sinkhacks.com/making-3d-printed-cat-armor/ #10yrsago Great moments in the history of black science fiction https://web.archive.org/web/20160308034421/http://www.fantasticstoriesoftheimagination.com/a-crash-course-in-the-history-of-black-science-fiction/ #1yrago Daniel Pinkwater's "Jules, Penny and the Rooster" https://pluralistic.net/2025/03/11/klong-you-are-a-pickle-2/#martian-space-potato Upcoming appearances (permalink) Barcelona: Enshittification with Simona Levi/Xnet (Llibreria Finestres), Mar 20 https://www.llibreriafinestres.com/evento/cory-doctorow/ Berkeley: Bioneers keynote, Mar 27 https://conference.bioneers.org/ Montreal: Bronfman Lecture (McGill) Apr 10 https://www.eventbrite.ca/e/artificial-intelligence-the-ultimate-disrupter-tickets-1982706623885 London: Resisting Big Tech Empires (LSBU) https://www.tickettailor.com/events/globaljusticenow/2042691 Berlin: Re:publica, May 18-20 https://re-publica.com/de/news/rp26-sprecher-cory-doctorow Berlin: Enshittification at Otherland Books, May 19 https://www.otherland-berlin.de/de/event-details/cory-doctorow.html Hay-on-Wye: HowTheLightGetsIn, May 22-25 https://howthelightgetsin.org/festivals/hay/big-ideas-2 Recent appearances (permalink) Launch for Cindy's Cohn's "Privacy's Defender" (City Lights) https://www.youtube.com/watch?v=WuVCm2PUalU Chicken Mating Harnesses (This Week in Tech) https://twit.tv/shows/this-week-in-tech/episodes/1074 The Virtual Jewel Box (U Utah) https://tanner.utah.edu/podcast/enshittification-cory-doctorow-matthew-potolsky/ Tanner Humanities Lecture (U Utah) https://www.youtube.com/watch?v=i6Yf1nSyekI The Lost Cause https://streets.mn/2026/03/02/book-club-the-lost-cause/ Latest books (permalink) "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 Reverse-Centaur's Guide to AI," a short book about being a better AI critic, Farrar, Straus and Giroux, June 2026 "Enshittification, Why Everything Suddenly Got Worse and What to Do About It" (the graphic novel), Firstsecond, 2026 "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, 2027 "The Memex Method," Farrar, Straus, Giroux, 2027 Colophon (permalink) Today's top sources: Currently writing: "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 (1081 words today, 48461 total) "The Reverse Centaur's Guide to AI," a short book for Farrar, Straus and Giroux about being an effective AI critic. LEGAL REVIEW AND COPYEDIT COMPLETE. "The Post-American Internet," a short book about internet policy in the age of Trumpism. PLANNING. 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/ https://twitter.com/doctorow 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