<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Y on Medium]]></title>
        <description><![CDATA[Stories by Y on Medium]]></description>
        <link>https://medium.com/@Nhlakanipho-Kubheka?source=rss-10c206cce7e7------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*H6SaDO4wZnNSKF5mkUCGJw.png</url>
            <title>Stories by Y on Medium</title>
            <link>https://medium.com/@Nhlakanipho-Kubheka?source=rss-10c206cce7e7------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 06:44:56 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@Nhlakanipho-Kubheka/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The AI Witch-Hunt: Publishing’s New Crisis of Authorship]]></title>
            <link>https://medium.com/@Nhlakanipho-Kubheka/the-ai-witch-hunt-publishings-new-crisis-of-authorship-42e915a6fe58?source=rss-10c206cce7e7------2</link>
            <guid isPermaLink="false">https://medium.com/p/42e915a6fe58</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[publishing-industry]]></category>
            <category><![CDATA[hachette]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[book-publishing]]></category>
            <dc:creator><![CDATA[Y]]></dc:creator>
            <pubDate>Wed, 14 Oct 2026 20:10:20 GMT</pubDate>
            <atom:updated>2026-10-14T20:10:20.884Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ts82wJjfmjjnFwObbuNXkA.png" /></figure><p>Article by: Nhlakanipho Kubheka</p><p>The publishing industry has a serious problem with artificial intelligence. It also has a serious problem with how it responds to that problem. Publishers are under pressure to protect intellectual property, preserve the value of human creativity and prevent undisclosed machine-generated content from entering their catalogues. Those concerns are legitimate. The increasingly casual treatment of AI allegations, however, threatens the careers of writers whose conduct has not been established through reliable evidence. An industry that depends on discovering original voices must decide whether it intends to evaluate manuscripts on their merits or require authors to defend their legitimacy before their work can be considered.</p><p>The controversy surrounding Nigerian author Jerry Falade illustrates the financial consequences. His debut crime novel, <em>Call Me, I’ll Hide the Body</em>, reportedly attracted an offer exceeding $2 million from Macmillan’s Minotaur imprint following a fourteen-way American publishing auction. Several substantial British offers were also reported. The book had generated considerable interest before questions emerged about the possible use of artificial intelligence during its production. In July 2026, Falade’s agents withdrew the manuscript from submission, explaining that they could no longer authenticate its development from conception to completion. The commercial opportunity was subsequently disrupted before the novel reached publication.</p><p>According to <em>The Guardian’s</em> report of 31 July 2026, Falade denies using AI to write his manuscript. His agents initially accepted his assurances but later said that aspects of his account had changed during a meeting on 29 July. They cited the need for confidence in his authorship when explaining their decision to end the professional relationship. These details matter. The case cannot responsibly be presented as proof that a publisher deliberately destroyed an innocent writer’s career, because the reported concerns about Falade’s account have not been independently resolved here. His denial must nevertheless receive the same serious consideration as the allegations against him. <a href="https://www.theguardian.com/books/2026/jul/31/crime-novel-deal-collapses-questions-ai-jerry-falade-call-me-ill-hide-the-body">The Guardian’s report</a> provides the principal public account of the dispute.</p><p>The central concern is the disparity between the uncertainty surrounding an allegation and the certainty of its consequences. A reported offer exceeding $2 million was associated with a manuscript that several publishing professionals had considered commercially promising. Once the authorship dispute emerged, the opportunity deteriorated rapidly. Whatever conclusion a future investigation might reach about the manuscript, the professional consequences had already begun. Falade lost representation, the acquisition process collapsed, and his name became associated with an allegation that he disputes.</p><p>The distinction between an allegation and an established finding is fundamental. Agents are entitled to expect honesty from their clients. Publishers are entitled to know what they are acquiring. An author who misrepresents substantial machine-generated work may breach contractual obligations and undermine reader confidence. None of these considerations removes the obligation to investigate claims carefully. When a decision can affect millions of dollars in prospective earnings and permanently alter an author’s reputation, the standard of evidence deserves particular attention.</p><p>The publishing industry’s present difficulty is that its methods for identifying AI-generated prose remain imperfect, while its appetite for taking action appears to be increasing. AI detectors examine textual characteristics associated with machine-generated writing. They do not provide a complete record of who wrote a manuscript, how it developed or which interventions occurred during editing. Their outputs can inform an investigation, but a detector score cannot independently establish authorship.</p><p>Research has documented the risks of false positives. A 2023 study published in <em>Patterns</em> tested seven AI-text detectors and found that the systems frequently misclassified essays written by non-native English speakers as AI-generated. The average false-positive rate among the tested essays was 61.3 per cent. The study concerned a particular collection of educational texts, and its findings should not be transferred directly to every detector or literary manuscript. Its significance is methodological: a detector can incorrectly classify authentic human writing, making independent verification essential before consequential decisions are made. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC10382961/">The study is available through the National Library of Medicine</a>.</p><p>The problem becomes more complicated when questions of literary style and cultural expression enter the discussion. Falade has argued that his prose draws on Yorùbá language and oral storytelling traditions, which use proverbs, parallel constructions, metaphor and elaborate phrasing. He believes that elements of this tradition contributed to the suspicion surrounding his work. His explanation does not settle every question about the manuscript, but it raises an important issue for an international publishing industry that frequently seeks distinctive voices from communities it does not necessarily understand in depth.</p><p>Literary expression varies across languages, regions and traditions. Writers develop through reading, education, cultural inheritance, imitation, experimentation and revision. Some favour restrained prose, while others employ repetition, rhetorical patterns and elaborate descriptions. A text that departs from familiar conventions is not inherently machine-generated. A writer’s improvement over time is not evidence of misconduct. Any assessment that relies heavily on stylistic expectations risks treating literary difference as grounds for suspicion.</p><p>This does not mean that accusations of AI use should be dismissed whenever an author offers a cultural explanation. Such an approach would be equally irresponsible. The requirement is evidence that can distinguish between competing explanations. Investigators should examine available drafts, revision histories, correspondence and other relevant material, while acknowledging that no individual item necessarily establishes the complete history of a manuscript. The purpose of an investigation should be to reach a defensible conclusion, not merely to find a justification for an initial suspicion.</p><p>The economic dimension of the debate becomes particularly significant when considering South African romance novelist and AI-publishing entrepreneur Coral Hart. Hart has publicly embraced the use of generative AI in commercial fiction and has promoted methods for incorporating the technology into publishing operations. Her approach differs fundamentally from Falade’s situation because her use of AI is an acknowledged component of her business model.</p><p>A February 2026 profile in <em>The New York Times</em>, titled “The New Fabio Is Claude,” described Hart’s experiment with AI-assisted romance publishing. Over approximately eight months, she established 21 pen names and published dozens of novels. Her work attracted attention because of the speed at which she was able to expand her output and the difficulties she encountered when attempting to generate convincing emotional and sexual intimacy through AI tools. <a href="https://archive.ph/1fwii">The profile is available here</a>.</p><p>Hart’s position deserves consideration on its own merits. She has approached AI as a commercial technology that can reduce the time required to produce books and expand the number of titles a publishing operation can release. Her work raises questions about literary quality, commercial efficiency and the future of independent publishing. Those questions should be addressed through criticism of the work and a clear understanding of how it was produced, rather than personal attacks on Hart.</p><p>The more difficult issue concerns the consequences of high-volume production for authors who continue to create their books through conventional writing processes. An author may spend a year developing a manuscript, revising its structure, correcting its prose and preparing it for publication. An AI-assisted operation can produce a considerably larger number of titles during the same period. The commercial results will vary, but the difference in production capacity has obvious implications for the availability of books, advertising competition and the difficulty of attracting readers.</p><p>Readers have limited time and disposable income. Publishing platforms must organise and recommend an expanding catalogue of material. Authors must compete for attention through reviews, publicity, advertising and audience engagement. Increasing production volume does not necessarily produce better books, and a larger catalogue does not guarantee that readers will discover more titles they enjoy. It can, however, increase competition for visibility and make the commercial environment more difficult for authors who have limited promotional resources.</p><p>The concern is particularly relevant to independent writers, including those in South Africa who are attempting to establish themselves without substantial financial support. Many already face difficulties securing representation, accessing professional editing, paying for cover design and reaching readers outside their immediate networks. The growth of AI-assisted publishing may introduce additional competition in an environment where discoverability is already a persistent challenge. Writers will have to compete for readers’ attention while the supply of commercially produced content continues to expand.</p><p>There is a separate question concerning the value of creative labour. If a publishing operation uses AI to generate substantial portions of a book, readers should have access to clear information about the nature of that contribution. Authors should understand the contractual implications of their production methods. Publishers should establish policies that distinguish between grammar correction, editorial assistance and substantial machine-generated content. These practices have different consequences for authorship and disclosure, and a responsible industry should not treat them as interchangeable.</p><p>The controversy surrounding Hart therefore raises questions about commercial production, whereas the Falade case raises questions about the reliability and fairness of authorship investigations. Both deserve serious scrutiny, but neither resolves the other. The presence of AI-assisted books in the market does not justify treating every writer as a potential offender. The possibility of false accusations does not remove the need to address undisclosed AI use. An effective publishing policy must accommodate both concerns.</p><p>There is also an uncomfortable question of institutional consistency. Publishers have begun incorporating AI into some operational functions, including marketing, publicity and internal communications. <em>Wired</em> reported in October 2026 that major publishing companies, including HarperCollins, Simon &amp; Schuster and Hachette, were expanding their use of AI tools amid staff concerns about transparency, labour and the effects on creative work. <a href="https://www.wired.com/story/book-publishers-are-quietly-using-more-ai-staff-are-revolting">The report was published by <em>Wired</em></a>.</p><p>Different applications of AI carry different risks, so publishers are not necessarily hypocritical merely because they use the technology internally while restricting its use in manuscripts. They must, however, explain their policies clearly. An organisation that permits AI-generated marketing copy while prohibiting substantial AI-generated fiction should identify the reasoning behind that distinction. Authors and employees should not have to discover the rules through inconsistent enforcement and public controversy.</p><p>The consequences of unreliable authorship assessments are becoming more apparent in literary competitions. On 14 October 2026, <em>The Guardian</em> reported that Zimbabwean writer Simbarashe Kundizeza had denied using AI to write <em>The Blue Fighter</em>, his Caine Prize-winning short story. The controversy followed an article in <em>The Atlantic</em> reporting that the detection software Pangram had classified much of the story as AI-generated. Kundizeza rejected the allegation, and the Caine Prize committee announced that it was reviewing the matter through what it described as a human-centred approach. The committee’s chair, Ellah Wakatama, questioned whether detection software should be allowed to determine authorship conclusively, particularly when Black writers’ work was being assessed. <a href="https://www.theguardian.com/technology/2026/oct/14/zimbabwean-author-simbarashe-kundizeza-denies-ai-story">The Guardian reported the dispute on 14 October 2026</a>.</p><p>This controversy does not establish that Kundizeza’s work was necessarily misclassified. The investigation remains important, and the available evidence must be assessed carefully. What it demonstrates is the difficulty of applying detection technology to literary work when the consequences include the possible loss of a major award and damage to a writer’s standing. The question of whether a story was generated by AI must be resolved through a process that considers evidence, context and the author’s response.</p><p>The recent controversy involving Haitian-Canadian novelist Thélyson Orélien further demonstrates why individual cases require careful treatment. In September 2026, Orélien’s novel <em>C’était ça ou mourir</em> was removed from the Goncourt prize longlist following AI allegations. On 13 October, Reuters reported that Orélien denied using AI to write the novel but acknowledged past plagiarism in earlier works. These are separate issues. His admission concerning earlier plagiarism is relevant to his record, but it does not independently establish that the novel under dispute was generated by AI. The distinction must be maintained if the publishing industry intends to assess individual claims accurately. <a href="https://www.reuters.com/world/haitian-canadian-author-thelyson-orelien-denies-use-ai-admits-past-plagiarism-2026-10-13/">Reuters’ report</a> documents his response and the separate plagiarism findings.</p><p>The pattern of disputes involving Black authors has also raised concerns about unequal treatment. Falade argued that major opportunities for several Black writers had been disrupted following AI allegations, while some white writers who had acknowledged using AI had continued to work without comparable consequences. These claims should not be accepted as proof of discriminatory intent without examining the circumstances of the individual cases. Nevertheless, they warrant scrutiny. When accusations repeatedly involve writers from underrepresented communities, publishing institutions should investigate whether their procedures apply consistently across different authors and literary traditions.</p><p>This question carries particular weight for African writers seeking international recognition. Publishing has a responsibility to respect linguistic diversity and cultural expression while maintaining credible standards for authorship. African writers should not have to produce additional proof of their legitimacy simply because their prose differs from the expectations of overseas editors or reviewers. Equally, cultural identity should not exempt anyone from a reasonable investigation into a specific and credible allegation. Fair treatment requires the same evidential standards for everyone.</p><p>The ripple effect may be most damaging for emerging writers who cannot afford representation, specialist advice or prolonged disputes. A prominent author may have access to professional support and an established audience. Someone seeking a first publishing contract may have neither. The suspicion alone can discourage an agent from pursuing a manuscript or a publisher from taking a commercial risk. An author may never learn why an opportunity disappeared, and there may be no public record of a decision influenced by concerns that were never properly investigated.</p><p>The industry also risks changing how writers approach their craft. If authors believe that unusual phrasing, elaborate prose or a significant improvement in style may trigger an accusation, some will begin making creative decisions according to what detection software is likely to accept. That pressure has no legitimate place in literary development. Authors should be encouraged to revise, experiment and produce distinctive work. Their primary concern should be the quality and integrity of their writing, not whether statistical software will misinterpret its characteristics.</p><p>Publishers must therefore establish procedures that protect both the integrity of their catalogues and the rights of the authors they represent. Their policies should specify which uses of AI are permitted, which require disclosure and which are prohibited. Contracts should explain the author’s obligations before a manuscript is acquired. Where credible concerns arise, the author should receive a clear account of the allegations and a reasonable opportunity to respond. Investigations should consider relevant documentation and expert assessment, and no single detection result should determine the outcome.</p><p>Confidentiality is also essential during an investigation. Public accusations can produce consequences before the facts have been established, and corrections may fail to restore the professional opportunities that were lost. Publishers, agents and prize committees should communicate carefully about unresolved cases. Where misconduct is established, the relevant parties should explain the findings and consequences. Where the evidence remains inconclusive, they should say so.</p><p>Authors can also protect their interests by retaining drafts, outlines, research notes and revision histories. These records may help demonstrate how a manuscript developed. However, the responsibility cannot rest entirely on individual writers to document every stage of their creative process. Writers use different methods, and the absence of extensive records should not automatically be interpreted as evidence of AI use. Publishers must develop reasonable procedures that recognise the limitations of the available evidence.</p><p>The industry must also confront its commercial incentives. AI-assisted production may reduce certain costs and increase output, but a rapid increase in available titles does not guarantee improvements in literary quality, reader satisfaction or working conditions. Publishers should examine whether their policies protect the value of creative labour and whether authors receive clear information about how technology is being used in the production and promotion of their work. These discussions are necessary even if they complicate the industry’s enthusiasm for efficiency.</p><p>There is no defensible reason to dismiss every AI allegation, nor is there a defensible reason to treat suspicion as proof. The distinction is straightforward, although the practical consequences are considerable. A publishing professional who identifies a credible concern should investigate it. An author who has used AI in a manner that violates a contractual requirement should be held accountable. An author accused without sufficient evidence should receive a fair assessment and protection from unjustified professional consequences.</p><p>Falade’s lost publishing opportunity, Hart’s experiments with AI-assisted production and Kundizeza’s current dispute represent different aspects of a rapidly changing industry. Together, they raise questions about authorship, commercial competition, cultural bias and the standards used to evaluate literary work. These questions will require clearer policies, more reliable evidence and a willingness to distinguish genuine misconduct from unresolved suspicion.</p><p>The publishing industry’s responsibility is not to preserve an imaginary past in which technology played no role in writing and production. Its responsibility is to protect the integrity of literary work, ensure that commercial arrangements are transparent and treat authors fairly. That requires acknowledging the legitimate concerns raised by generative AI without allowing the technology to become an excuse for careless professional decisions.</p><p>An author should not be guaranteed a publishing deal merely because they claim to have written a manuscript. Publishers must be able to investigate credible concerns and enforce their contractual standards. But an author should not lose a career-defining opportunity because a detector produces a concerning score or because a style appears unfamiliar to the people evaluating it. Serious allegations require serious evidence, and the consequences of an investigation must reflect the strength of its findings.</p><p>The AI witch-hunt is a professional crisis because the costs of getting an authorship decision wrong are substantial. A publishing opportunity may disappear before an investigation reaches a defensible conclusion. An award may be withdrawn. A writer’s reputation may be damaged beyond repair. The commercial consequences are immediate, while an eventual correction may offer little practical relief.</p><p>The industry needs to make a choice about the standards it intends to uphold. It can establish clear rules for AI use, investigate alleged violations carefully and protect writers from unsubstantiated accusations. Or it can allow uncertainty to dictate decisions, with emerging authors bearing the greatest financial and professional risks. The latter approach may reduce the immediate anxiety of publishers and agents, but it will weaken trust in the institutions responsible for discovering and supporting new writing.</p><p>Publishing depends on authors who are willing to produce work that deserves attention. It also depends on readers who trust the claims made about that work. Protecting both requires honesty about AI, consistency in enforcement and restraint when the facts remain disputed. The industry cannot claim to defend human creativity while creating conditions in which writers must fear the consequences of demonstrating it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=42e915a6fe58" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>