Three recent controversies reveal more about the current state of the publishing industry than they do about the inherent quality of large language model (LLM)-generated writing.
Since 2012, the distinguished British literary magazine, Granta, has consistently published the regional winners of the annual Commonwealth Short Story Prize. This year, however, one of the selections for this esteemed award raised eyebrows, as it appeared to have been authored by artificial intelligence.
Jamir Nazir’s story, “The Serpent in the Grove,” exhibited numerous characteristics commonly associated with LLM-generated prose, such as mixed metaphors, anaphora, and recurring lists of three. (The author notes the irony of this observation being presented as a list of three, assuring readers that this piece, like all their work, was written without AI assistance.) Initially, the author expressed skepticism regarding the claims that Nazir’s story was AI-generated. While acknowledging the increasing use of LLMs for writing assistance or full generation, a cautious approach to the growing "AI paranoia" among peers was maintained. Alleged AI "tells" often cited include em dashes, the word "delve," lists, and short, impactful sentences punctuating longer ones.
Yet, as a human writer, the author has undeniably employed all these stylistic elements previously. LLMs, after all, are trained on vast datasets of human writing, thus mirroring the patterns they consume. Despite this mimicry, AI-generated prose often possesses an unsettling quality, a subtle "offness" that is hard to pinpoint immediately. If these supposed AI indicators are present in human writing, it raises the fundamental question of how one can definitively ascertain human authorship.
Nabeel S. Qureshi, a former visiting scholar of AI at the Mercatus Center at George Mason University, was among the first to flag the suspected AI involvement in Nazir’s narrative. For Qureshi, the opening two sentences alone provided sufficient evidence:
“They say the grove still hums at noon. Not the bees’ neat industry or the clean rasp of cutlass on vibe, but a belly sound — as if the earth swallows a shout and holds it there.”
Qureshi communicated via email, stating, “In general, AI writing has a particular rhythm that I’ve learned to pick up on which is hard to describe. There’s a spectrum from ‘AI helped me edit’ to ‘AI wrote this’ — this case reads to me like the latter end of that, though of course I don’t know for sure.”
The core issue lies in the lack of absolute certainty, even when AI use is strongly suspected. Razmi Farook, director-general of the Commonwealth Foundation, acknowledged the allegations concerning AI in prizewinning stories, including Nazir’s. Farook clarified that all entrants are required to confirm their submission is original, unpublished work, and all shortlisted writers have personally affirmed that no AI was used in drafting their stories.
Farook emphasized, “Until a sufficient tool or process to reliably detect the use of AI emerges that can also grapple with the challenges pertaining to working with unpublished fiction, the Foundation and the Commonwealth Short Story Prize must operate on the principle of trust.”
Granta’s response involved running Nazir’s story through Claude and "asking whether it was AI-generated," as publisher Sigrid Rausing explained. Claude’s lengthy reply concluded that it was “almost certainly not produced unaided by a human.” However, Claude is a chatbot powered by a large language model, not a dedicated AI detection tool. While AI tools often surpass human readers (especially literary judges) in identifying LLM-produced prose, Granta’s approach suggested a misunderstanding of AI functionality, implying they sought an answer directly from the AI source itself. Rausing conceded, “It may be that the judges have now awarded a prize to an instance of AI plagiarism — we don’t yet know, and perhaps we never will know.”
Publications are increasingly falling victim to AI-generated submissions, some even attributed to non-existent authors. Initial suspicions arose that Nazir himself might be a fabrication, though Kevin Jared Hosein, a previous Commonwealth Short Story Prize winner, verified Nazir’s identity and shared recent correspondence about the AI allegations. Nazir also published a poetry collection in 2018. Nazir did not respond to The Verge’s request for comment. In a related incident in March, Hachette withdrew Mia Ballard’s horror novel, Shy Girl, following accusations of AI use, which Ballard denied, blaming a hired editor.
Beyond outright generation, a critical question emerges regarding the acceptable boundaries of AI use for authors and journalists. While full LLM-generated prose is widely deemed unacceptable, what about leveraging AI for idea generation, research, or transcription services? At what point does reliance on these tools dilute the authorship of the work? This week, Polish Nobel laureate Olga Tokarczuk admitted to using AI in her creative process – a use case at the other end of Qureshi’s spectrum, yet one that alarmed many admirers of the Nobel Prize winner.
Tokarczuk, who received literature’s highest honor in 2018, elaborated, “I often simply throw into the machine an idea with the prompt: ‘Darling, how could we beautifully elaborate this?’ Even though I know about its hallucinations and numerous factual errors in the fields of quantitative economics or factual data, I have to admit that in the fluid field of literary fiction, this technology is an asset with unbelievable leverage. At the same time, I feel an acute human grief over an era that is disappearing never to return. I’m heartbroken by the departure of traditional literature written in isolation over months, a work conceived in the mind of a single conscious individual. In all of this, I’m damn mournful for Balzac, Cioran, and the inimitable Nabokov, because in spite of my enthusiasm, I don’t believe that any modern chat has managed to speak in their exquisite manner.” These remarks, delivered in Polish at an event in Poznań, unfortunately went viral concurrently with the Commonwealth Prize scandal. (Her comments were translated into English by a human.) However, her stance on AI is far more nuanced than initial headlines suggested. Tokarczuk later issued a three-point statement to Lit Hub, clarifying that she does not use AI to write her forthcoming book, but rather for “faster documenting and checking of facts,” always independently verifying information. She added, “I am sometimes inspired by dreams, but before this sentence too is cornered and torn to pieces by the experts, I hasten to report that they are my own dreams.”
The strong reaction to Tokarczuk’s initial comments, and her subsequent need for clarification, underscores a broader, arguably justified, apprehension within the publishing world regarding AI. While LLM-generated prose might become commonplace, the question remains whether this is a desired future. Thousands threatened to boycott Barnes & Noble after CEO James Daunt stated he had no issue selling AI-written books, provided they carried disclaimers of non-human authorship. Daunt later partially retracted his comments, telling the Los Angeles Times, “Book banning is a clear and present danger, so we are very careful with demands to ban any books,” while also committing “not to sell AI generated books that masquerade to be by real authors.”
None of these developments, however, fully elucidate the uncanny quality of AI-generated work, or how to differentiate poorly written LLM prose from poorly written human prose. When Nazir’s story was analyzed using Pangram, an AI and plagiarism detection software, it registered as 100 percent AI-generated. Pangram identified Nazir’s use of triads, the word “stubborn” (six times more frequent in AI text), and the phrase “as if it had” (five times more frequent) as key indicators. Yet, this paragraph, penned by a human, also contains a list of three.
Still unsatisfied, the author submitted an unpublished excerpt from their forthcoming, currently-being-edited book to Pangram. A single paragraph contained two triads. (Admittedly, it was not a strong section, hence the editing.) Pangram nevertheless declared the excerpt 100 percent human-written, which was accurate, but left the author still seeking clarity. A subsequent, stronger excerpt yielded the same result. The first chapter of Verge editor Kevin Nguyen’s novel, Mỹ Documents, also passed Pangram’s test. Interestingly, Pangram itself analyzed all Commonwealth Prize winners and found that two of the 2026 awardees, along with the 2025 winner, showed signs of AI generation. Ultimately, both human-produced and AI-generated work possess a certain ineffable quality. Perhaps AI prose is akin to obscenity: you recognize it when you encounter it, even if you cannot fully articulate why.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.