The short story The Blue Fighter was announced as the winner of the Caine prize for African writers on 6 October. Within five days, a journalist claimed the piece had been flagged by an AI-detection system as almost entirely machine-generated, igniting a controversy that quickly spread across literary circles.
How the accusation originated
The claim originated in an article written by Vauhini Vara, who ran the story through Pangram a tool that itself relies on artificial intelligence to spot patterns typical of AI-written text. According to Vara, the software marked the manuscript as largely artificial, with the only human element being an epigraph attributed to South African poet Vangile Gantsho. Vara described the prose as saturated with the “not x, but y” constructions, an overabundance of metaphors, and “platitudinous aphorisms”, concluding, “Taste is subjective, but trust me, this story is bad – maudlin and full of tired tropes.”
Author’s rebuttal and the prize’s response
Simbarashe Kundizeza took to his X account to categorically deny the allegation, stating, “For the avoidance of doubt, I wrote ‘The Blue Fighter’. I did not use generative AI to write it. It is entirely my own work.” He added that he would continue collaborating with the Caine prize team and keep writing. The prize’s administration confirmed that it had opened an internal review, emphasizing a “human-centred approach” and noting that the chair, Ellah Wakatama, questioned the reliability of detection software. Wakatama told BBC’s Front Row that using such tools as a “final arbiter” is premature, especially when it risks unfairly targeting black writers.
Broader context and similar controversies
Kundizeza’s case is not isolated. In July, Trinidadian author Jamir Nazir faced similar accusations after winning the Commonwealth Foundation’s short-story prize, while a Canadian-Haitian novelist, Thélyson Orélien, saw his book removed from the Goncourt longlist after an anonymous claim of AI authorship. These incidents illustrate a growing tension between emerging detection technologies and the literary community’s confidence in authorial authenticity.
Technical debates surrounding detection tools
Vara referenced Pangram’s own research, which suggests false-positives are rare, even for non-native English speakers, and cited a peer-reviewed study highlighting the tool’s “strong accuracy” on mock academic papers. Nevertheless, she admitted that independent third-party evaluations remain limited. Critics argue that the current generation of AI-detection models can misinterpret creative stylistic choices—such as the “not x, but y” pattern—as algorithmic fingerprints, potentially penalising writers who experiment with language.
What the dispute reveals about literary craft
In his statement, Kundizeza challenged the notion that a writer cannot evolve. He said, “To suggest that a writer cannot write in different styles or genres, or improve his writing substantially after five or more years fundamentally misunderstands the craft of writing.” The episode underscores a larger conversation: as AI-assisted tools become more sophisticated, the literary world must balance vigilance against genuine fraud with an appreciation for artistic growth and diversity of voice.
For now, the Caine prize’s investigation continues, and Kundizeza remains firm that his work is wholly human. The outcome may set a precedent for how literary institutions handle AI-related allegations in the future, shaping the intersection of technology and creative expression.
