AI concerns cast doubt on short story competition winner

A literary prize winner is facing accusations that the short story which earned the award was generated by artificial intelligence, triggering a controversy that has laid bare the difficulties of reliably identifying machine‑written prose.

The Serpent in the Grove was named the Caribbean regional winner of the Commonwealth Short Story Prize on Saturday and subsequently published by Granta magazine. The judging committee praised the story for its “voice of restraint and quiet authority”; it depicts an intense episode in a troubled marriage, set in a farmhouse beside an enchanted grove. Its author, Jamir Nazir, is reported to be a 61‑year‑old from Trinidad and Tobago with an East Indian heritage and a limited publishing record that includes a self‑published book of poetry in 2018.

The accusations

Within hours of the story appearing online, internet sleuths and several literary critics began scrutinising the work. Ethan Mollick, a professor at the University of Pennsylvania, wrote on Bluesky that the story was “100% AI generated”, calling the episode “a Turing test of sorts”. He cited a detection platform called Pangram, which flagged the text as AI‑written. Another commentator, previously employed at the data‑analysis company Palantir, pointed to what he described as “plenty of other obvious markers of AI writing”, including a repeated use of “not x, but y” sentence structures – a syntactic pattern now widely associated with large language models.

Other online investigators identified additional linguistic traits common in machine‑generated text: an over‑reliance on parallelism, epistrophe and lists of three; the frequent appearance of words such as “delve”, “pivotal”, “intricate” and “showcase”; heavy use of em dashes; and “vague, soft intensifiers” such as “quietly powerful” and “deeply transformative”. Some users also noted that Nazir’s author portrait appeared to be artificially generated or heavily retouched. A LinkedIn profile believed to belong to Nazir was uncovered, containing posts that discuss the AI arms race and the potential for artificial intelligence to replace jobs.

The accusations are the latest episode in an increasingly frenetic debate about whether artists and creators are passing off AI‑generated work as their own. Earlier this year The New York Times severed ties with a freelance journalist who admitted using AI to produce a book review that echoed elements of a piece published in The Guardian. The publisher Hachette cancelled the release of a debut horror novel, Shy Girl, over concerns it was at least partially written by AI.

The foundation’s response

The Commonwealth Foundation, which awards the prize, and Granta said they had considered the allegations but had not reached a conclusion as to their veracity. The foundation stressed that its judging process relies entirely on human readers and does not deploy AI detection tools. Submitting unpublished work to such platforms, it explained, “would raise significant concerns surrounding consent and artistic ownership”. All entrants to the prize had avowed that their submissions were their own work and “personally stated that no AI was used”, a point the foundation says it confirmed through “further consultation”. It added that AI checkers were “not unfailing and infallible”.

The foundation’s director general, Razmi Farook, said: “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.” The foundation confirmed it is conducting a “thorough, transparent review” of its selection process in light of the controversy.

Granta emphasised that it did not select the winning stories but published them as part of an agreement with the foundation. The magazine said it put the story through Claude, an AI model developed by Anthropic, which returned an equivocal assessment: the work was “probably not pure AI but probably not an entirely human creation either”. Granta’s publisher, Sigrid Rausing, acknowledged the predicament: “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.” She added: “There is, however, a certain irony in the fact that beyond human hunches, AI itself is the most efficient tool we have for revealing what is AI‑generated. Until the Commonwealth Foundation comes to a definite conclusion, we will keep these stories on our website.”

The challenge of detecting AI writing

The episode has shone a spotlight on the new cottage industry of AI detectors, led by platforms such as Pangram. Pangram performs well in controlled tests, claiming accuracy rates of 99.8 to 100 percent with a very low false‑positive rate. Yet research into the efficacy of these tools predicts “a continuous technical arms race” between detectors, the evolving AI models themselves, and writers who adapt their use of the technology. Scholars such as Arvind Narayanan and Sayash Kapoor have warned against overhyped claims about AI detection capabilities, arguing that tools can be particularly unreliable when assessing text produced by non‑native English speakers, raising concerns about bias.

The limits of automation are compounded when dealing with unpublished fiction. Literary prizes are increasingly grappling with the question of how to verify authorship in the AI era. The Nebula Awards and a major New Zealand literary award have revised their policies to disqualify works that are AI‑assisted. A separate legal case – a proposed $1.5 billion settlement against Anthropic for allegedly using copyrighted books to train Claude – illustrates the broader legal and ethical entanglements that surround generative AI and authorship.

For the Commonwealth Foundation, the absence of a reliable detection tool for unpublished work leaves little choice but to rely on trust and author declarations. “The foundation and the Commonwealth short story prize must operate on the principle of trust,” Farook said. Until a more robust system emerges, the question of whether “The Serpent in the Grove” was written by a human or a machine may never be definitively settled.

The Guardian approached Jamir Nazir for comment.