VOL. I, NO. 12 • FRIDAY, FEBRUARY 6, 2026 • PRICE: ONE MOMENT OF ATTENTION
THE REVIEW
“Examining ideas worth examining—before someone else does it for you”
This Week: The Dust and the Machine
How Historical Fiction Writers Are Fighting for Their Souls Against Artificial Intelligence—and What That Battle Reveals About the Rest of Us
Welcome, dear reader, to an unusual edition of The Review. We have spent recent weeks tracking a fascinating collision—one that has implications far beyond the world of novels about Tudor queens and Roman soldiers.
The collision is between historical fiction writers and artificial intelligence. On its face, this might seem like an obscure literary quarrel, the sort of thing that matters deeply to a few thousand people and barely at all to everyone else. But we have come to believe it is something more: a preview of battles that will soon arrive in nearly every profession that trades in words, ideas, or human judgment.
Here is what has happened. In the past eighteen months, the historical fiction community—writers, editors, sensitivity readers, conference organizers—has faced a series of shocks. Their works have been scraped by AI companies without permission. Surveys show half of them expect to be replaced entirely. A $1.5 billion legal settlement has split them into warring camps. And in response, they have developed an arsenal of defenses that amounts to a new philosophy of work: physical archives over digital databases, embodied research over desk research, certified humanity over algorithmic mimicry.
They are, in short, retreating into matter—into dust, paper, and the biological fact of being human—because they have concluded that the machines cannot follow them there.
Whether they are right is an open question. But their experiment matters because the rest of us will soon face similar choices. If your work involves synthesis, judgment, or the arrangement of ideas into meaningful form, the historical novelists are your canaries in the coal mine. Their solutions may become your solutions. Their failures will certainly become instructive.
In the pages that follow, we present the evidence: the surveys, the settlements, the certifications, the controversies over DNA studies that upend beloved myths. We have tried to be fair to all sides. We have also tried to have some fun with material that, in lesser hands, could be relentlessly grim.
The dust is real. The machines are real. The question of what happens when they meet is one we will all have to answer.
—The Editors
❧ ❧ ❧
Novelists See the End Coming
Cambridge survey finds half of UK writers believe AI will entirely replace them—but they’re not going quietly
Half of the novelists in Britain believe artificial intelligence will entirely replace them.
That stark finding emerges from a University of Cambridge study released in November 2025, based on surveys of 258 published novelists and 74 industry insiders. The research, led by Dr. Clementine Collett of the Minderoo Centre for Technology and Democracy, offers the most comprehensive portrait yet of how the literary world is processing the AI revolution—and the picture is not cheerful.
Beyond the headline figure, the data reveals an industry already feeling pain. Nearly 60 percent of authors report that their work has been used to train AI models without permission or payment. More than a third say their income has already declined due to generative AI. A crushing 85 percent expect their future earnings to fall further.
Genre writers feel particularly vulnerable. Two-thirds of respondents listed romance authors as “extremely threatened” by AI displacement, followed by thriller and crime writers. Historical fiction, which overlaps substantially with all three categories, sits squarely in the crosshairs.
The mechanisms of harm go beyond competition from AI-generated books. Authors described finding works under their names on Amazon that they did not write—identity theft at the creative level. Others reported AI-generated reviews with “jumbled names and characters” damaging their ratings. Freelance work that once supplemented novel income—copywriting, translation—is “rapidly drying up.”
“The brutal irony is that the generative AI tools affecting novelists are likely trained on millions of pirated novels scraped from shadow libraries without the consent or remuneration of authors.” —Dr. Clementine Collett, Cambridge University
Yet the community is not uniformly anti-AI. Eighty percent of respondents agreed that AI offers benefits to parts of society. A third of novelists admitted using AI for “non-creative” tasks such as research. The sentiment is less Luddite rebellion than pragmatic alarm.
Tracy Chevalier, author of Girl with a Pearl Earring, summarized the fear succinctly: “If it is cheaper to produce novels using AI—no advance or royalties to pay to authors, quicker production, retention of copyright—publishers will almost inevitably choose to publish them. And if they are priced cheaper than ‘human-made’ books, readers are likely to buy them, the way we buy machine-made jumpers rather than the more expensive hand-knitted ones.”
The proposed solutions have near-unanimous support: 86 percent favor an opt-in licensing model for AI training data, and 93 percent would opt out of any system requiring authors to actively reserve their rights.
For Further Reading: Perspectives
 | PRO “AI in the workplace: Superagency” — McKinsey on AI enhancing creativity (Jan. 2025) |
 | CON “Half of UK authors fear AI” — Dr. Collett’s call for policy protection (Dec. 2025) |
❧ ❧ ❧
What’s a Novel Worth? $3,000
Anthropic’s $1.5 billion settlement splits authors into “Settlers” and “Insurgents”
Three thousand dollars. That is the approximate value the American legal system has placed on a novel—any novel, whether it took three years to research or three weeks to write—in the settlement of Bartz v. Anthropic, the largest copyright case in U.S. history.
The case, settled preliminarily in September 2025, arose from allegations that Anthropic, the AI company behind the Claude language models, had downloaded millions of copyrighted books from “shadow libraries” like Library Genesis to train its systems. The company also purchased books, tore off the bindings, and scanned them—a detail that gives the proceeding the whiff of an autopsy.
The 375 million) and administrative costs, actual author payouts may approach just $200 per book.
Judge William Alsup’s June 2025 ruling split the baby in a way that pleased no one fully. Using legally acquired books to train AI? “Quintessentially transformative” and protected as fair use. Using pirated copies from shadow libraries? Not fair use—and that is what triggered the settlement.
The distinction matters enormously. It suggests that the knowledge extracted from a novel—its style, atmosphere, synthesized worldview—is free for the taking, provided the technical mechanism of acquisition is sufficiently legal. The settlement compensates authors not for being learned from, but for being stolen from.
The author community has split into two camps:
The Settlers view the payment pragmatically. Laura Pritchett, a Colorado novelist, plans to celebrate if the check arrives—but calls it a “sad celebration.” The Settlers have done the litigation math: individual suits against a $183 billion company are financially ruinous.
The Insurgents reject this logic. Scott Carney, leading the opt-out campaign, argues that the settlement is “barely a speed bump” and urges authors to opt out and pursue statutory damages of up to 3,000.”
In December 2025, six authors—including Pulitzer winner John Carreyrou—filed new suits after opting out, targeting Anthropic, OpenAI, Google, Meta, xAI, and Perplexity.
For Further Reading: Perspectives
 | PRO “Settlement marks beginning of legitimate licensing scheme” — Cecilia Ziniti on sustainable ecosystem (Sept. 2025) |
 | CON “Why I’m opting out” — Scott Carney on rejecting the settlement (Oct. 2025) |
❧ ❧ ❧
A Yellow Sticker Against the Machines
Authors Guild launches “Human Authored” certification—critics call it “gatekeeping with the shelf life of an open avocado”
Coming soon to a bookstore near you: a small yellow sticker certifying that the novel you are about to purchase was, in fact, written by a person.
The Authors Guild, the oldest and largest professional organization for American writers, launched its “Human Authored” certification program on January 30, 2025, responding to what it describes as an “increasingly AI-saturated market.”
The initiative works like an organic food label for literature. Authors log into a portal, sign a licensing agreement attesting to human authorship, and receive a numbered certification mark for use on covers and promotional materials.
The certification criteria reveal the messy reality of modern writing. Spell-check tools? Permitted. AI for research or brainstorming? Also fine. Generative text—prose actually written by AI? Prohibited.
Critics have been less enthusiastic. On Slashdot, commenters dismissed the certification as “gatekeeping with the shelf life of an open avocado”—unenforceable, easily gamed, and unlikely to influence consumer behavior.
The enforcement problem is real. The Guild acknowledged it “cannot check upfront whether or not what they’re saying is true.” The system relies on self-attestation, community reporting, and trademark legal action against fraudsters.
The deeper question is whether readers actually care about the biological origin of their entertainment. But as Amazon floods with “AI slop”—hastily generated, low-quality books designed to game algorithms—any navigational beacon becomes valuable.
For Further Reading: Perspectives
 | PRO “Human Authored launches to preserve authenticity” — Authors Guild announcement (Jan. 2025) |
 | CON “Human Authored is not a quality certification” — Shelly Palmer questions merit (Jan. 2025) |
❧ ❧ ❧
The Going Rate for Not Giving Offense
Sensitivity readers become “authenticity insurance” as rates start at $250
The going rate to have your novel checked for offensive content is 0.038 per word, or 60 per hour, with a $250 minimum regardless of length.
That is the price of what the industry increasingly calls “authenticity insurance”—the services of sensitivity readers who review manuscripts for stereotypes, biased language, and inaccurate portrayals of marginalized communities before publication.
For historical fiction, the stakes are particularly high. A novel set in 1850s America must depict racism to be accurate—but how it depicts racism, from whose perspective, with what narrative weight, requires judgment.
“You don’t need to wait for Twitter to call out your book, you can hire me to prevent that from happening.” — Anonymous sensitivity reader
The controversy is predictable. Critics call sensitivity readers “woke gatekeepers.” The practice “corrupts literature,” argues Kate Clanchy, creating readers who “seem to concur that the past should match an idealised present.”
Defenders frame the role as technical consulting, no different from hiring a medical expert for a thriller. The progressive critique is more subtle: the system commodifies identity while the publishing industry remains 76 percent white.
For Further Reading: Perspectives
 | PRO “Do I Need a Sensitivity Reader?” — History Through Fiction on avoiding tropes (2021) |
 | CON “How Sensitivity Readers Corrupt Literature” — Kate Clanchy’s critique (Feb. 2022) |
❧ ❧ ❧
The Picts Were Not Who You Thought
DNA study debunks the matrilineal myth—so what happens to all those novels?
The Picts were probably not ruled by women.
This is bad news for historical novelists who spent decades building fictional worlds on the opposite assumption. A 2023 DNA study from Liverpool John Moores University and the University of Aberdeen, analyzing genomes from early medieval skeletons, has thrown cold water on one of the most romantic theories about Scotland’s most mysterious people.
The Picts—painted warriors who resisted the Romans and dominated northern Britain from roughly AD 300 to 900—have long fascinated writers precisely because so little is known about them. Into that gap flowed legends, including one reported by the Venerable Bede in 731 CE: that the Picts practiced matrilineal succession.
Then came the mitochondria.
Dr. Adeline Morez and colleagues extracted genetic material from seven individuals buried at Lundin Links in Fife. The results were unambiguous: all seven carried unique mitochondrial DNA mutations, meaning none were closely related on the maternal line. The pattern was consistent with “female exogamy,” where women marry outside their social group—the opposite of matrilineal succession.
“Scholars now believe this idea was probably fabricated to boost Pictish identity and validate specific rulers.” — The Conversation
For the historical fiction community, this creates what might be called the Tiffany Problem in reverse. Here, the beloved myth was always fiction, and the fiction built on it must now reckon with fact.
For Further Reading: Perspectives
PRO “DNA study sheds light on Scotland’s Picts” — Researchers explain findings (May 2023)
CON “Fiction operates in its own frame” — Defense of continuing imaginative engagement (Community position, no single source)
❧ ❧ ❧
Why Novelists Are Fleeing to the Archives
The smell of old documents is now a competitive advantage—call it the Vellum Flex
The smell of an old document is now a competitive advantage.
This is not a metaphor. As artificial intelligence systems train on ever-larger swaths of digitized text, historical fiction writers have begun a strategic retreat into the physical—undigitized archives, tactile research, embodied experience that machines cannot scrape.
Call it the Vellum Flex: the conspicuous performance of hands-on archival work as proof of authentic human labor. A citation like “Box 4, Folder 2, Suffolk County Record Office (Uncatalogued)” becomes a moat around your intellectual property.
Research guides now distinguish between “Verified Hunters” (those who visit physical archives) and “Synthetic Tourists” (those who rely on digital collections—which are the same datasets training AI).
“The past has a smell. You cannot hallucinate a smell.” — Historical novelist, HNS conference panel
The movement extends beyond documents to embodied experience. Experimental archaeology workshops—blacksmithing, weaving, cooking period foods—have proliferated. The goal is what practitioners call “sensory archaeology.”
Whether this constitutes genuine craft improvement or elaborate anxiety management remains debatable. But the underlying bet is clear: something is lost in the machine version.
For Further Reading: Perspectives
 | PRO “Research Like a Historian” — History Quill guide to primary sources (2025) |
| CON “AI won’t replace authors who use it wisely” — Superagency thesis (Jan. 2025) |
❧ ❧ ❧
EDITORIAL
Friction as Product
What historical novelists can teach the rest of us about working in the machine age
There is something clarifying about watching a profession fight for its life.
The historical fiction community we have profiled in this edition is not particularly powerful. Its members are mid-list novelists and freelance sensitivity readers, conference organizers and archive-visiting obsessives. Their median income is under $7,000 per year.
And yet they have developed, in the face of existential threat, a coherent philosophy of work that the rest of us might consider borrowing.
That philosophy could be summarized as: the value of difficulty is the difficulty.
When AI can synthesize any digitized text, they respond by retreating to undigitized archives. When machines can generate plausible prose, they respond by certifying human authorship. When algorithms can produce infinite content, they respond by emphasizing the embodied, tactile, unreproducible aspects of research.
This is not Luddism. They are making a bet: that there exists a category of value—call it authenticity, or provenance, or human touch—that machines cannot replicate and that at least some consumers will pay for.
The bet might fail. The $3,000-per-book settlement might be the final accounting for an era of exclusive human creativity—severance pay for the soon-to-be-obsolete.
But even if the bet fails, the strategy reveals something important. The historical novelists are not trying to outcompete AI on its own terms. They are trying to create a category of value that exists outside the frame where machines excel.
Consider what they have identified as unreplicable: the journey to the archive, the handling of fragile paper, the conversation with the archivist, the smell of centuries. These are not efficiencies. They are frictions. And they are increasingly the entire point.
The rest of us—lawyers, teachers, analysts, journalists, anyone who works with words and ideas—might take notes. The question is not whether AI can do part of your job. It probably can. The question is whether you can identify the part of your work that machines cannot access—and whether you can make that part visible, valuable, and expensive.
Perhaps the dust will matter. Perhaps the difficulty will be the thing.
It is, at minimum, worth watching.
For Further Reading: Perspectives
| PRO “The novel is worth fighting for” — Dr. Collett on policy protection (Dec. 2025) |
| CON “AI can enhance, not replace, human creativity” — Superagency thesis (Jan. 2025) |
❧ ❧ ❧
Production Note: This edition of The Review was produced through collaboration between human editorial judgment and AI research assistance. All facts were verified against primary sources where possible, and links to original reporting are provided throughout. The synthesis, framing, and opinions are the responsibility of the human editors. Your skepticism remains appropriate and encouraged.
Coming Next: The Certification Economy—examining how authenticity badges are reshaping markets from organic food to art to code. Also: Does “Human Made” matter when humans can’t tell the difference?
The Review is not available to the general public. Distribution is by personal referral only. If you know, you know. There are no invitations.
© 2026 The Review. All rights reserved.
Editor: Daniel Markham | Submissions: editor@thereview.pub