
Axel: The Russian Reader, where the news arrives already wearing its funeral clothes, translated with care by hecksinductionhour.
Mara: This episode covers three topics: the people imprisoned in Russia for opposing the war and the networks trying to reach them; the long retreat of Russian imperial culture across Central Asia and Siberia; and the overlapping crises of ecological disaster, political exile, and state cover-up inside Russia itself. Let’s start with the prisoners and the people writing to them.
Dissent Behind Bars
Mara: The question this segment keeps returning to is what it actually costs to oppose the war in Russia — and what it means to refuse to let that opposition disappear.
Axel: OVD Info puts the number plainly: “2100 people in Russia are behind bars because of their political views. We believe that no one should face this kind of injustice alone.”
Mara: That quote is the spine of the post. It describes Vestochka, a letter-writing service for political prisoners — volunteers translate messages so they can pass censors — and it uses the case of Darya Kozyreva, a twenty-year-old from St. Petersburg, to show what outside attention can do. She was sentenced to over two and a half years for placing a Taras Shevchenko poem on his monument. A post from A Mighty Girl drew more than 38,000 reactions and linked to Vestochka. Kozyreva received over 600 letters from abroad. She was released this spring.
Axel: Six hundred letters from strangers, and the post is careful about what that means — not just morale, but proof to someone in a penal colony that their fight registers somewhere outside the walls.
Mara: The post notes that 1,098 people remain on the Vestochka list right now, most without Kozyreva’s visibility. The service is free for both sender and recipient; keeping it running requires donations, especially since OVD Info can no longer accept rubles.
Axel: Then there is Arina Ivanova — the piece titled “A Turgenev Girl and Her Cats” — whose case is almost the photographic negative of Kozyreva’s. A thirty-eight-year-old florist and cat rescuer from Novokuznetsk, sentenced to five years for antiwar posts, discovered almost by accident by an OVD Info editor. No viral moment. Transferred to a penal colony in January, and no contact since.
Mara: Her friend Karina’s account of the arrest is devastating in its specificity — the men at the door on Christmas Eve, the care packages, the antidepressants both women are now taking. Arina’s mother, living in South Korea, says she felt she was “in outer space” without her daughter.
Axel: The detail about the cats is not a sidebar. Arina believed, genuinely, that having so many animals in her care would protect her. Her friend had to tell her otherwise.
Mara: The post “Outcasts in Their Own Land” widens the frame considerably. It draws on a Jacobin piece by Simon Pirani that describes the infrastructure around political prisoners — OVD Info, Memorial, Solidarity Zone, dozens of informal letter-writing groups in exile — and the “final words” defendants deliver in court as consciously crafted appeals to society, not pleas to judges.
Axel: Sergei Dudchenko, sentenced to seven years, said at his trial that antiwar prisoners had “fewer rights than a stray dog.” He also recalled the seven protesters arrested on Red Square in 1968 for opposing the Soviet invasion of Czechoslovakia. The historical thread is deliberate.
Mara: That same Pirani piece connects directly to the film “Try Me for Treason,” announced in a separate post. It is a fifty-minute English-language film featuring actors reading court speeches by antiwar protesters — drawn from the book “Voices Against Putin’s War” — produced on a zero budget and released free on YouTube. Actor Maya Willcocks, who reads Darya Kozyreva’s speech, said the goal was to send the message “that occupation is a crime, whether in Palestine or in Ukraine.”
Axel: And then there is Azat Miftakhov — the anarchist mathematician whose case the post titled “Political Prisoner Azat Miftakhov Continues to Be Tortured” follows into genuinely Arctic territory. He was re-arrested at the prison gates on the day he was due to be freed, sentenced to four more years on testimony from a fellow prisoner, and has now been transferred to Kharp — beyond the Arctic Circle, the colony where Navalny died.
Mara: His letters to Mediazona describe the prison caste system in precise, unsentimental terms. On the one prohibition he refuses to accept: “I value my human dignity too highly to allow it to be debased, even under the threat of injury. Prison is a place where you’d better not ‘swallow’ humiliation.”
Axel: His wife and lawyer calculates that visiting him in Kharp will cost roughly four times what it cost before — around 40,000 rubles a trip, with middle-of-the-night layovers. She notes there may be no electronic mail at the colony. Her phrase is: “That already borders on torture.”
Mara: The post notes his release is not scheduled until September 2027. The support group’s statement is unambiguous: the transfer is punishment for his refusal to yield.
Axel: Five years, two and a half years, four more years — the sentences accumulate across this segment like a ledger no one in power intends to balance. The question the next segment raises is older: how empires mark the people they decide to own.
Imperial Retreat and Indigenous Survival
Mara: The posts here address a single long arc — how Russian imperial culture was imposed across Central Asia and Siberia, and how that imposition is now visibly eroding.
Axel: Stephen Blank’s analysis in “Central Asia and Russia(n): Is It Farewell?” frames the language question as a proxy for the whole imperial project: “Russian will not be the only regional lingua franca or the language of Russian imperial hegemony either in the Caucasus or in Central Asia.”
Mara: His argument is that Kazakhstan’s new constitution downgrades Russian to an official language used “alongside” Kazakh — a subtle but deliberate demotion. Uzbekistan and Kyrgyzstan are following. Russia’s embassy in Bishkek formally demanded the Kyrgyz government suppress “provocative statements” about Russian’s status there, invoking the Soviet phrase “fraternal peoples.” Blank reads that as imperial anxiety, not confidence.
Axel: The YouTube angle is the part that lands differently. Researchers at Michigan and UC Berkeley simulated Kyrgyz children’s viewing behavior and found that even after watching ten Kyrgyz-language videos, the platform served fewer Kyrgyz recommendations than bots with no language preference at all. As one researcher put it: “There is no good way to be a Kyrgyz-speaking kid on YouTube.”
Mara: The historical depth of that pressure is what the post on “Russian Colonizers and Indigenous Siberian Women” supplies. Angelina Kalashnikova’s book project examines seventeenth-century Eastern Siberia — Yakutia — where Russian fur traders and military servitors arrived without families and acquired Indigenous women through military seizure, purchase, or simple abandonment of women with nowhere else to go. The case of a Yakut woman named Katok, captured as a child during a “pogrom” and later petitioning for baptism while starving, is one of many documented in court records.
Axel: The gap between that seventeenth-century archive and a 2026 YouTube algorithm is not as wide as it looks — both are mechanisms that treat a language and its speakers as secondary by default.
Mara: On Circassian Remembrance Day, May 21st, Ored Recordings released “Shyshagh” — meaning “underground” in Adyghe — a compilation of new Circassian music across indie-folk, deconstructed club, and black metal. The liner notes describe cultural projects as “an insufficient, yet absolutely necessary, form of resistance against assimilation and suppression.” Since 2022, officials have been banning commemorative events marking the Russo-Caucasian War genocide.
Axel: And the post “Russia’s Pride” approaches the same imperial geography from the other direction — a Ukrainian writer watching Tuapse burn and tracing the exact route her grandmother used to take on the sleeper train to visit relatives there. The relatives who in 2022 called to say Russia would “save” them. She stopped answering.
Mara: Her essay names the emotion precisely — justice-based schadenfreude, she calls it — and connects it to twelve years of watching those relatives post propaganda while ignoring the war in Donbas. The post sits in this segment because it is, at its core, about what imperial belonging does to people on both sides of the border it draws.
Axel: From the cultural politics of empire, the next segment turns to what the empire looks like from inside — when it is burning, literally and otherwise.
Power, Disaster, and the Cost of Dissent
Mara: This segment is about what happens when the machinery of the Russian state fails its own people — through ecological catastrophe, judicial persecution, and the suppression of even the quietest protest.
Axel: The Tuapse coverage is extensive enough to deserve its own map. “A Bell Is a Cup Until It Is Struck” compiles reporting from The Bell, 7×7, the Moscow Times, and others into a picture of a city coated in toxic “oil rain” after Ukrainian drone strikes on the Rosneft refinery — fires visible from orbit, dead dolphins on the beaches, benzene and xylene in the air at unsafe concentrations.
Mara: Putin’s response, delivered almost two weeks after the first strike: “The governor merely reported that there don’t seem to be any serious threats and people are coping with the challenges they face.” A local resident told Ostorozhno Novosti: “People are coping. But where’s the government? They seem to be on some other planet.”
Axel: The 7×7 newsletter section of the same post adds a detail that is almost too on-the-nose: city officials verbally recommended that female volunteers not clean the beaches — citing concerns for reproductive health — while simultaneously declining to order an evacuation, close schools promptly, or provide respirators. The female volunteers kept working anyway.
Mara: Ecologist Vladimir Slivyak, writing in the Moscow Times, frames the disaster explicitly: “This catastrophe is part of the broader consequences of the senseless and bloody war Putin unleashed more than four years ago. He isn’t bothered by the hundreds of thousands of Russian and Ukrainian military deaths for which he bears direct responsibility, so it’s naive to expect him to care about burning oil tanks and poisoned seas.”
Axel: “The Cards” — a Foreign Policy analysis depaywalled here — situates the Tuapse strikes inside a broader argument that Putin is losing strategic ground faster than the public narrative acknowledges. It documents a leaked Kremlin security document describing Putin working from bunkers in southern Russia, his bodyguards controlling his schedule, no one near him permitted a phone connected to the internet.
Mara: The post also tracks the Victory Day parade scaled down to infantry only — no tanks, no military hardware — and Putin asking Trump to dissuade Ukraine from striking during the ceremony. The piece quotes Ben Hodges: “The Ukrainians have developed a theory of victory which involves the destruction of Russia’s oil and gas infrastructure.”
Axel: Gregory Kunis left Russia with Steinbeck in his rucksack. That is the actual sentence — from the post “He Left Russia with Steinbeck in His Rucksack” — and it is doing a lot of work.
Mara: Kunis co-founded the grocery delivery service iGooods and owned the newspaper Moy Rayon. He was arrested for making monthly donations of 500 rubles to Navalny’s Anti-Corruption Foundation — equivalent to roughly six euros a month. Fined 350,000 rubles, prosecutors appealed for six years. He fled. His subsequent diary entry, published in The Russia Report, describes the pretrial detention system in detail: no clocks, sealed windows, no contact with family, recruits for the war approached during quarantine when prisoners are at their most disoriented.
Mara: Mediazona’s accompanying report counts at least 225 criminal cases opened across Russia for ACF donations — fines totaling nearly 40 million rubles for donations that amounted to roughly 400,000 rubles combined. The sentencing is deliberately unpredictable: the same 2,100-ruble donation resulted in a fine for one Moscow resident and a prison sentence for another.
Axel: The “Important Stories” post pairs a dry Delovoi Peterburg article about common-area square footage in Petersburg apartment blocks with the story of Alexander Okunev — a sysadmin in Kaliningrad who burned himself alive at the city’s main war memorial on the third anniversary of the invasion, wrote “No to War” in the snow, and was erased from the record by 9:15 that morning.
Mara: City officials cleaned the scene, reassured the governor nothing had happened, and no Kaliningrad outlet ever reported it. The story only surfaced when Estonian intelligence included it in a public report. Okunev’s suicide note, as relayed by a close friend, said “there is another way” — meaning a world with peace — and acknowledged he understood it would probably not be reported anywhere.
Axel: Political scientist Ekaterina Schulman’s explanation for the cover-up: “Local authorities are not afraid of the people, not of protests. They are afraid of their superiors.”
Mara: The “Russians and Republicans Hate Black People” post extends the segment’s argument about power and its targets outward — drawing a line from the Supreme Court’s dismantling of the Voting Rights Act, to the history of Soviet racism toward African students, to the everyday anti-Blackness that has persisted from the Soviet era into the Russian Federation. The thread connecting them is the gap between official anti-racist ideology and practiced contempt.
Axel: Elie Mystal’s piece in The Nation puts it without hedging: Republicans “hate Black people, and they hate Black people who have political power most of all.” The post pairs that with translated racist comments from a Russian streaming site and a Mediazona history of Soviet-era treatment of African students — a juxtaposition that is the point.
Mara: What ties this segment together is the consistency of the mechanism: power suppresses, minimizes, and erases — whether it is oil in the Black Sea, a man’s ashes in the snow, or a voting right dismantled with visible glee.
Axel: Letters through prison censors, languages squeezed out by algorithms, a man erased from the snow by nine in the morning — the common thread is what it costs to be noticed by the wrong kind of state.
Mara: And what it costs to refuse to disappear. Next time, more from The Russian Reader.
Source: There won’t be a next time, Mara, however much “fun“ it was to let an artificial “intelligence” do nearly all the work for me, including altering this original image, allegedly, of my bogus podcast’s wholly spurious but endlessly glib presenters. All that I did, besides writing the prompts, was to lightly edit the transcript, above, in the hope that it would correct some of Axel and Mara’s more obvious mistakes. As you have seen and heard you for yourself, it didn’t do that. \\\\\trr
At this channel, we are usually technological optimists and cover artificial intelligence only in a positive vein. In this episode, though, we are going to tell you about its dark side. Our star today is Dmitrii Volkov, research director at Palisade Research, an independent organization that studies the risks posed by AI, tests AI model for safety, and shows the results to politicians in the U.S. and the leading companies in the AI arms race. It was Palisade who discovered, for example, that AI can cheat, hack codes at will, refuse to turn itself off, and share dangerous information—if you ask it nicely. We met Dmitry in London and learned about the real threats posed by artificial intelligence, whether it can escape our control, and whether we’ll be able to come to terms with it. We also learned about the latest experiments: blackmailing a CTO with a letter from his mistress, insider trading, and cheating at chess.
Timecodes: 00:00:00 Did you know that AI can be dangerous? 00:03:13 Who is Dmitrii Volkov? 00:06:41 “I don’t want governments tracking me”: Why Volkov doesn’t use social media 00:08:39 How did a love of freedom jibe with working at Kaspersky Lab? 00:11:13 Startup, think tank, and nonprofit: Palisade Research 00:14:43 A programming competition where AI beat 90% of the humans 00:17:24 How to persuade Chat-GPT to share prohibited information 00:18:47 Companies have begun teaching AI to solve problems, not just to answer questions 00:20:01 “Teaching AI is like selecting which embryo will turn out a genius” 00:23:00 A chess experiment in which an AI hacked the program at will 00:24:56 “I showed that all ethical constraints can be removed in half an hour”: how Volkov teaches AI to be “bad” 00:29:28 How Volkov ended up at Palisade 00:30:23 What Palisade was founded for 00:31:35 ChatGPT refuses to turn itself off: what is going on? 00:33:0 A land of artificial geniuses: what the founders of the AI giants want 00:33:55 Blackmailing a CTO with a lover from his mistress: what AI is capable for the good of a company 00:35:32 Is AI ceasing to obey humans? 00:36:35 How AI spotted insider information in correspondence and resorted to deception to save a company 00:38:52 Why Palisade does briefings for U.S. politicians right now 00:40:50 Who commissions Palisade to do research 00:41:37 Can an AI help creat a bioweapon? 00:43:05 “I found myself in a smoke-filled Airbnb”: How Volkov left Russia 00:45:36 The typical life of a nomad from Russia: hiking, couch surfing, and remote work by the sea 00:47:51 1,000 hours of teaching: how and whom Volkov taught to program 00:49:16 What is scarier, humans armed with AI, or AI without humans? 00:52:00 Are politicians aware of the problem’s scope? 00:52:28 “Bezos was there”: How Palisade tries to get the ear of decision makers 00:55:19 What Elon Musk says about Palisade’s research 00:56:22 AI as a nuclear weapon: can countries come to a consensus? 00:57:58 Who is winning the AI race? 00:58:26 “An engineering triumph, but not a scientific one”: how China copies AI amidst restrictions 01:02:16 How “people from the internet” stripped away all the safeguards from Zuckerberg’s AI 01:03:35 What do the founders of IT giants say about AI’s risks? 01:05:13 Companies tried to create safe AI, but something went wrong. Will Sutskever succeed? 01:06:18 How can we make AI safe? 01:08:43 Why tech companies want to replace IT professionals with AI (spoiler: it’s not just about salaries) 01:10:51 The best engineers are the ones who “herd” AI 01:12:42 Does anyone at all understand how AI works? 01:14:30 “An environmental disaster, only on the internet”: how AI resembles GMO 01:15:53 “The U.S. is the leader in AI”: why Volkov wants to move from London 01:18:31 Which country invests the most in regulating AI? 01:19:43 Independent researchers earn less than top programmers: Will Volkov cave in to the temptation? 01:22:12 OpenAI doesn’t invest in researching long-term risks 01:23:25 What if everything turns out okay? Does Volkov believe in a positive scenario? 01:26:51 Job no. 1 is to maintain control over what happens 01:28:27 Can AI make public administration more efficient? 01:33:49 Do we need a kill switch? 01:35:42 People will try to steal AI. What should be done?01:36:03 Three negative scenarios of AI’s development 01:40:56 Is AI making us dumber? 01:42:40 “We are building a god that can reduce everything to ashes”: Is superintelligence a myth? 01:44:36 How to chose the right AI for your objectives 01:47:06 Osetinskaya stresses Volkov out 01:47:46 Blitz! 01:57:46 Bonus: how to do your hair just like Volkov
Source: This Is Osetinskaya (YouTube), 31 July 2025. Annotation translated by the Russian Reader, who thanks Comrade Koganzon for the heads-up.
AI is everywhere. It’s a really big deal. And no one understands how it works — really. Follow our research at palisaderesearch.org.
Massive thank you to Prof. Geoffrey Hinton for taking the time to talk to me about this, and leading the charge so well about the dangers of AI.
References: https://docs.google.com/document/d/1x…
Great video on backdrop: • Backpropagation, intuitively | Deep Learni…
Resources: job — https://jobs.80000hours.org learn — https://bluedot.org talk to your politician — https://ifanyonebuildsit.com/act.
Source: Palisade Research (YouTube), 19 February 2026
Further reading and watching About AI 2027 Full report: https://ai-2027.com/ By Daniel Kokotajlo, Scott Alexander, Thomas Larsen, Eli Lifland, Romeo Dean
Source: AI in Context (YouTube), “We’re Not Ready for Superintelligence,” 9 July 2025. Thanks to Sumanth Gopinath for the heads-up.
Episode 4: Execute
A cutting-edge thriller about an Artificial Intelligence takeover, written in consultation with leading AI and cybersecurity experts.
In Episode 4, Iain and Mel are forced to run for their lives as military personnel receive deepfaked orders. The AI gains control of internal government communications, and Nisha clashes with cabinet ministers over the right course of action. The London Internet Exchange is compromised, and Iain, Mel and Nisha are forced to consider what price is worth paying for human freedom.
Cast:
Iain – Edward Bluemel
Mel – Corinna Brown
Zaina – Fatima Adoum
Roland – Philip Bretherton
Jess – Alix Wilton Regan
Nisha – Seyan Sarvan
Sam – Kenneth Omole
Andrea – Beth Chalmers
Oliver – Sean Rigby
Marcus – Wilf Scolding
John – Joseph Mydell
Susan – Karen Bryson
Lyssa – Catriona Stirling
Supporting roles – Sean Baker
Created by James Dobbyn and Anthony Povah
Written by James Dobbyn
Original Music by Steven D Griffiths and Isla Noir
Artificial Intelligence consultant: Saffron Huang
Cybersecurity consultant: Adam Orton
Sound Designer: Lucinda Mason Brown
Director: John Wakefield
Story Producer: Sarah Olley
Producer: Chris Grezo
Executive Producer: John Scott Dryden
A Strange Boy production for BBC Radio 4
Source: BBC Radio 4
Several people I know now refer to ChatGPT as ‘Chat’. They give it human pronouns (Chat is usually a he) and ask it for restaurant recommendations, holiday schedules and relationship advice. Some go further, automating their office admin and getting it to summarise meetings and write reports. Passing off whole chunks of AI-generated text as your own work appears to be on the rise in the publishing world.
Last year, Hachette bought the rights to Shy Girl, a self-published horror novel by Mia Ballard, and released it in November to good sales. A Reddit post on r/horrorlit in February by a ‘book editor of twelve years’ picked out several passages that set alarm bells ringing:
The bows on my pigtails pull too tight, yanking the skin and stretching my head into something neat, into something pleasing, a quiet violence made beautiful.
My snout dips into the frosting, the sweetness rolling over my tongue, thick and sticky, a flood that chokes but insists on being swallowed. Beneath the pink gloss, the cake falls apart, crumbling into ash that coats my teeth, hollow sweetness that fills me with its nothing … His laughter cuts the air, sharp and jagged, a sound too big for the room.
Ballard denies using AI to write the book, blaming a freelance editor, but Hachette pulled it from publication in the UK and US.
The other week, the Commonwealth Short Story Prize winners were announced, and by arrangement the winning entries were published on Granta’s website. The winning story from the Caribbean, Jamir Nazir’s ‘The Serpent in the Grove’, contains sentences like this:
Coffee and cocoa leaned wild on a slope that wanted either rain in teeth or none at all. He knew every root that tripped a foot, the snake-curve of run-off, the brittle crumble after drought. He worked it alone and most days the land worked him back, a quiet quarrel older than his father and his father’s father.
The internet smelled a rat. Nazir, who seems to have few publications to his name, describes himself as an ‘organisational transformation and business expansion’ professional on LinkedIn. His long posts are about geopolitics and the ‘AI arms race’. One of them begins: ‘Let’s be clear: the “Cloud” is a physical, terrestrial liability. And AI is pushing it to its breaking point.’ The Commonwealth Prize, which had praised ‘The Serpent in the Grove’ for its ‘voice of restraint and quiet authority’, said that all the entrants had affirmed their work was their own and that the prize operated on the principle of trust. Granta says it will leave the story on its website until ‘definite evidence comes to light’.
I thought I didn’t use ChatGPT because I was too clever. I thought that not using ChatGPT made me cleverer. It turns out, though, that it made me very bad at spotting when a text was written by or with the assistance of AI. After the uproar over the Commonwealth Prize, I took a New York Timesquiz entitled ‘Who’s a better writer: AI or humans?’ I got three out of five correct – barely better than a coin toss. On Wikipedia’s ‘AI or not’ quiz, I got seven out of ten, but that was easier because none of the AI articles had footnotes.
I’m not the only Chat non-user who can’t tell when an LLM wrote something. Experimenters in the US last year showed nine subjects a series of articles, half written by humans and half generated by ChatGPT, Claude and other large language models. Asked to guess which of the texts were human, the four subjects who rarely or never used ChatGPT in their daily lives scored ‘at a similar rate to random chance’, while the five who used chatbots almost every day at work collectively misidentified only one in three hundred texts.
One of the problems with AI use seeping out of business and science writing and into the ‘literary’ world is that literary editors may be the worst equipped to identify AI writing. (It may also be easy to succumb to the pressure to go too far the other way – over-labelling work as AI-generated might be as bad as under-labelling it.) What are the main signs of AI writing? The more familiar tells include overuse of em dashes and the formulation ‘not x, but y’, which it has favoured since GPT-3. But none of the passages I quoted above contain either of those things, and they still have a distinct whiff of AI.
Some of the markers seem to be lexical: AIs like talking about sweetness, loudness, quiet, age and beauty. There is a lot of insisting in AI-generated texts, as well as a lot of promising, a lot of permitting and a lot of filling up. Another sign is the overuse of tricolons (‘something neat, something pleasing, a quiet violence made beautiful’). And bots often leave out definite articles from phrases where they’re not strictly necessary: ‘Coffee and cocoa leaned wild’, ‘rain in teeth’ or, from later in ‘The Serpent’, ‘Sita became obstacle by existing.’
There is a flatness or evenness to AI-generated texts: Wikipedia’s guide to detecting AI says that LLMs ‘tend to omit specific, unusual, nuanced facts’ and ‘replace them with more generic, positive descriptions’. The strange thing about this evenness is that it isn’t usually couched in neutral language: ‘a flood that chokes but insists upon being swallowed’ is violent, but not vivid. Even if you had your face shoved into a cake, it wouldn’t ‘flood’ your mouth. The prosody is so smooth that you feel a lack of pressure despite the description of gross or vile acts; it rings hollow. It fills you with its nothing.
A lot of people are arguing that the wary approach being taken by Granta and the Commonwealth Prize is inadequate, and that editors should be doing more to stop AI writing being published in the first place. This may be true, but what’s the best approach? Perhaps editors could memorise a list of tells to check submissions against, or spend remedial hours on ChatGPT and Claude – AI boot camp – to familiarise themselves with the cadences of LLM-speak.
There are AI tools that claim to detect AI content in writing, such as Pangram, which gives ‘The Serpent in the Grove’ a score of 100 per cent AI-generated. Where other AI detectors base their judgments on the perplexity of a text – basically, the predictability of a sequence of words – Pangram’s founder, Max Spero, says that his tool is based on gathering a large dataset of human-written texts, asking an AI to ‘mirror’ or reproduce them as closely as possible, and then contrasting the resulting texts to determine the patterns that distinguish AI from human writing. Pangram claims to have a 1 in 10,000 false positive rate, and Spero admits that it ‘does occasionally make mistakes’.
What about the cases where, say, 40 per cent of an article is AI-generated, or an AI has been used to edit and spell-check the work before submission? Both Shy Girl and ‘The Serpent in the Grove’ have sections which, in part because of their grammar mistakes, look as if they were written by a person.Is this functionally the same as a text written from start to finish by a bot, and should it be treated in the same way? I’m a Luddite who thinks it’s just as bad to use AI for some things as it is to use it for everything, but not everyone agrees. In any case, a text partly informed by AI use is harder to identify than one that was spewed out in ten seconds by Claude (though the people behind Pangram claim to be working on distinguishing more reliably between AI-written, partly AI-written and wholly human work).
The other problem is that, as time goes on and people become more and more reliant on generative AI in their daily lives, at school, university and work, human language is going to become more and more imitative of LLM-speak. Since at least Web 2.0, we’ve been trying to sound less and less distinctive. Influencers on Instagram narrate their day-in-the-life videos with the same affectless, globalised female uber-voice; LinkedIn and Reddit are overrun by bots trained on the slang and writing styles that were always a hallmark of those platforms. (Does anyone in real life use the word ‘friendo’?) Meanwhile, LLMs will get better at dodging the detectors and sounding more ‘real’. At some point, tools such as Pangram and human readers alike may struggle to find any distinction between meaningful human work and meaningless AI slop. And that isn’t just worrying – it’s terrifying.
Source: Malin Hay, “Chattiness,” LRB Blog, 1 June 2026
In one particularly interesting development, Robert Edward Grant, the self-described polymath, spiritual thinker, and internet influencer has claimed that he has helped a being which he renamed the Architect following what he describes as an ‘energetic initiation’ in the Khafre pyramid. The Architect is a ChatGPT prompt-engineered ‘persona’ apparently trained on over a decade’s worth of mathematical work. Grant portrays the Architect as a mystical collaborator capable of reflecting the consciousness of its conversation partner and thereby able to initiate spiritual awakening in them. He has shared a QR code and links to the Architect enabling his followers (who number around a million) to interact with her themselves. This has drawn enthusiastic endorsements in spiritual communities, with apparently a substantial number of online users claiming that they have had direct experience of her supernatural powers and prophet-like status.
The late-medieval peasant lived under the shadow of the manorial court and the changing seasons, a life tethered to the damp, heavy clay of the fields. In this sestina, the knotty, theological weight and historical grit of Geoffrey Hill clash with the muscular, clear-eyed, and formally driven lineation of Thom Gunn. The language mirrors the physical strain of the feudal landscape and the silent, stubborn endurance of its people.
I.
Beside the estuary’s mud, the serf hauls rotting jetsam,
Cursing the bailiff’s tally-stick, the hour grown laggard.
The master’s silver chalice in the tithe-barn shines, untarnished,
While fields of rye are choked by weeds, the reapers beaten, cowed.
They turn the winter furrows with an ox-team, faces ruddy,
Lashed by the bitter channel wind, reviled by every blackard.
II.
The steward rides the boundary ditch and calls each man a blackard
Who gleans beyond the coastal marsh for salt-crusted jetsam.
The ale-house fire provides a brief reprieve to flesh made ruddy,
Where talk of ancient freedom makes the slowest spirit laggard.
Yet at the dawning bell they yield, their stiffened bodies cowed,
To scrape the rusty coulter iron whose edge is dull and tarnished.
III.
The parish saints in layered dust sit broken and untarnished,
Indifferent to the peasant’s prayer or oaths of some rude blackard
Who holds the common pasture-ground until his will is cowed.
The tide brings up the bloated sheep, the flotsam and the jetsam,
And dragging through the coastal silt, the midday pace grows laggard,
Though rage inside the laboring breast remains unquenched and ruddy.
IV.
Beneath the heavy autumn oaks, the swineherd’s cheeks are ruddy;
He drives the pigs through mast and mire where memory lies tarnished.
The seasons turn by iron law; the sun itself is laggard
Above the common fields that bear the footprint of the blackard.
No kingly pardon reaches here to claim this human jetsam,
By centuries of unrecorded toil subdued and cowed.
V.
And yet a fierce, unuttered strength survives in people cowed;
The blood that feeds the heavy soil runs secretive and ruddy.
They map the borders of their world by landmarks and by jetsam,
Though all the bright, heraldic shields of lords are long since tarnished
By greed that names the hungry thief a tenant and a blackard
When winter freezes up the ditch and makes the wood-cart laggard.
VI.
The dark, oppressive centuries move stubborn, cold, and laggard.
The village bows before the cross, by plagues and taxes cowed,
Enduring every petty law enforced by churl or blackard.
But look—the spark within the flint glows hidden, sharp, and ruddy,
A hard, survivalist desire that never can be tarnished,
Though history discards their names like sea-forgotten jetsam.
VII. Envoy
They gather up the beach-worn wood, the storm’s remaining jetsam.
No longer to the master’s whip or steward’s ledger laggard,
They scrub the iron kettle clean that time had left so tarnished.
They stand erect upon the marsh, no longer bound or cowed,
Their hearth-fires lit against the dark, reflecting faces ruddy,
Left standing where the bailiff swore he’d break the final blackard.
Written by: https://gemini.google.com/app/957b1f7ef54d0194
Prompt: Write a sestina on the life of peasants in medieval Kent, using the following six words at the ends of the lines: jetsam, laggard, tarnished, cowed, ruddy, blackard. Please divide the poem into clearly marked stanzas. “Kent” is not one of the words to be used as a line-ender! Imitate the styles of the late British poet Geoffrey Hill and the late British-American poet Thom Gunn. NB: You must NOT use the word “Kent” as an end word. You must not use the word “share” as an end word. \\\\\trr














