Because literary journals tend to have small circulations, few readers are likely to have come across “The Whistleblower,” a short story by the Zimbabwean writer Simbarashe Steyn Kundizeza. Published in 2021 in a journal called Transition, the story is a taut and closely observed political thriller in miniature. Its narrator, a sharp but inexperienced computer-science graduate in Harare, Zimbabwe, encounters an intelligence officer at the mall—a former teacher of his—who now wants something from him. At first, the graduate is nervous for what turns out to be the wrong reason: “I’d had a brief relationship with his niece Agatha and I wasn’t sure whether he knew about this or how he felt about it if he did.”
A more recent story with Kundizeza’s name on it, “The Blue Fighter,” will surely find its way to more readers. Last week, it was named the winner of the prestigious Caine Prize for African Writing, which awards 10,000 British pounds (about $13,000) for a short story by an African writer. Some previous winners, including Namwali Serpell and NoViolet Bulawayo, are now among the world’s most celebrated authors.
Taste is subjective, but trust me, this story is bad—maudlin and full of tired tropes. Here, the Zimbabwean narrator is the son of a hapless father who squanders the family’s money trying to breed fish in tanks, to the mother’s chagrin. “‘We can’t afford this house anymore,’ she said one evening, placing the notice in front of him while he tested the seal on his newest creation,” the story reads. “Not with anger, but with the quiet resignation of someone who had known all along this moment would come.” At the end, the narrator compares the last remaining fish to his family: “Not broken, just learning how to live differently. Alone, sometimes. But still moving. Still here.”
That this narrator has all the personality of a chatbot is probably not coincidental. The AI detector Pangram flags almost the entire story as AI-generated; a notable exception is its epigraph, also about a father with a fish fixation, which is excerpted from what’s described as an unpublished poem by Vangile Gantsho. This isn’t the first prize-winning story to be flagged by Pangram. It comes after similar scandals over this year’s Commonwealth Prize and Prix Goncourt—two other renowned literature contests—in which Pangram flagged AI use among finalists.
In the case of the Commonwealth Prize, one of them eventually won; the Prix Goncourt pulled a writer from consideration after suspected AI use was compounded by plagiarism allegations over his earlier work. Any of these incidents might have been written off as a fluke. Together, they point to something of a crisis of trust—in both writers and technology—that threatens to undermine a global system of literary recognition.
[Read: The literary AI scandal that changes everything]
AI detection is fallible, which has raised concerns among some readers and writers about the prospect of false positives. But Pangram, which uses AI for its classification of text, is highly accurate: The company says its tool mistakenly labels human text as AI in about one in 24,000 cases; outside research also finds false positives to be extremely rare. When I ran all 25 of the previous Caine Prize-winning stories through Pangram, it labeled them as 100 percent human-written; the same is true of Kundizeza’s “The Whistleblower.”
I reached out to Kundizeza about Pangram’s assessment of “The Blue Fighter” through his agent, Adrienne Ingrum at the Serendipity Literary Agency. She wrote that Regina Brooks, the agency’s CEO, would send me “evidence that the author and we at Serendipity are completely confident that the story is Simba’s own work.” Neither Ingrum nor Brooks followed up before publication.
I also spoke at length with Ellah Wakatama, who chairs the Caine Prize’s board of trustees. Wakatama, a longtime book editor, told me that AI-generated fiction is not eligible for the prize. As soon as she learned of an inquiry I’d sent about the Pangram assessment of Kundizeza’s story, she and her colleagues initiated an investigation and contacted the author. Wakatama told me that she and a colleague will interview the prize judges, and that she’ll reread Kundizeza’s story and talk with him in detail about his methods. She added that a colleague has been in contact with Kundizeza’s agent about whether a verification process might include, say, reviewing previous drafts.
Yet Wakatama forcefully rejected the notion that an AI detector could, or should, be used to gauge AI involvement. “My stance, as the person who leads the prize, is that if there is such an accusation, I want proof that does not come from a machine.” She said she planned to approach Kundizeza with “compassion” and the presumption of honesty. Wakatama is also Zimbabwean. “As an African,” she said, “my first stance is, if I ask you, Are you telling the truth? and if you say yes, I am going to believe you—especially because the accusers are not coming from amongst our own people.”
Wakatama remarked on details in Kundizeza’s story that strike her as so typically Zimbabwean that a Zimbabwean must have written them: the shame of moving from a nice suburb to a less privileged township; how the mother’s initial forbearance with the father gives way to taking matters into her own hands. Wakatama also perceives multiple layers of meaning. She doesn’t think AI models are capable of that kind of cultural or literary nuance.
[Read: The Pangram backlash unfolding on college campuses]
But then, she acknowledged that she doesn’t use AI or AI detectors. “My first response when somebody asks me to use AI is that I think about water and energy,” she said—that is, AI companies’ excessive use of both. “That honestly overwhelms me, and I don’t want to.”
Wakatama isn’t involved in choosing the Caine Prize winner—a panel of judges reads all the submissions—but her approach to AI is common among organizations that award prizes or other honors to writers. Partly, this is a pragmatic response to the risk of false positives. “Our legal parameters don’t allow us to contest the honesty of our writers just because a tool says that,” Razmi Farook, the director-general of the Commonwealth Foundation, told me during the scandal over its prize.
But this wariness also reflects a general aversion, in literary and arts circles, to engaging with AI products, even if just to understand them better—including Pangram. The start-up’s product itself is based on AI, and it belongs to the Silicon Valley ecosystem; the company recently raised $50 million in funding, according to a person familiar with the deal.
The skepticism is easy to understand, but it might be shortsighted. Research suggests that the people most familiar with AI output excel at sniffing it out. I’ve covered AI for almost a decade, and have written fiction for even longer; in short, I’ve had plenty of exposure to both literature and chatbot output. I know from experience that, for example, AI models aren’t terrible at picking up on cultural details.
When I read “The Blue Fighter,” I was struck by the proliferation of language that is common in AI-produced writing: the not x, but y sentence constructions; the glut of metaphors; the platitudinous aphorisms. “Anger burns out eventually,” the narrator’s mother says at one point. “Hunger doesn’t.” The earlier story by Kundizeza lacked those stylistic qualities, its prose and dialogue far more plainspoken and natural-sounding: “So how are things with my niece, huh?” the intelligence officer asks at one point.
What I also noticed, more disturbingly, was the presence of stereotypes about African life common among outsiders: a blundering, unemployed father; a resilient matriarch who dispenses clichéd approximations of wisdom; an environment of deprivation. Research consistently shows that AI products reproduce cultural tropes, including those having to do with the African continent.
[Read: The most famous AI writing tic is also the most mysterious]
The novelist and critic Tope Folarin, a former Caine Prize winner who has written about how dominant white culture shapes global literature, told me after reading the story that it would be unfortunate if Kundizeza had relied on AI for any of it, because its plot is “deeply human” and “deserves human expression.” He also said, “There are a lot of writers from the global South who feel this extreme pressure to write to expectations so their work is legible to the West, and obviously chatbots have glommed onto that.” He called the incident a “tragedy.”
Wakatama, for her part, explained her suspicion of AI detection by referencing the long, fraught history of science being used to oppress people of color. To a skeptical reader, her invocation of race might sound like a preemptive excuse. But in our conversation, she came across as open-minded and intellectually honest. I considered her wariness of Silicon Valley well founded and reasonable—particularly because the writers implicated in three major literary scandals this year were all nonwhite and non-American.
Pangram’s internal research finds false positives to be rare or nonexistent across multiple races and languages, as well as for English-language learners writing in English. In a separate peer-reviewed study, researchers at the Vrije Universiteit Brussel found zero false positives for English-language writing by non-native English speakers. It’s also notable that Pangram didn’t detect AI in any previous Caine Prize–winning stories or in a pre-ChatGPT story of Kundizeza’s. Yet third-party research into Pangram’s effectiveness for writers across nationalities is limited.
Still, it seems that arts communities’ revulsion toward AI has fostered a self-defeating underinvestment in understanding how exactly it functions, and how those who are repelled by AI companies’ presence in cultural spheres might beat back its encroachment. If an AI detection tool is known to work really well, couldn’t we at least use it alongside other methods? Barring that, couldn’t we build some expertise in detecting fakery the old-fashioned way—through human training and intuition?
“Language carries culture,” the Kenyan novelist Ngũgĩ wa Thiong’o once wrote, “and culture carries, particularly through orature and literature, the entire body of values by which we come to perceive ourselves and our place in the world.” Language, he added, is “thus inseparable from ourselves as a community of human beings with a specific form and character, a specific history, a specific relationship to the world.” Ngũgĩ, following this line of reasoning, chose to return to his native language of Gĩkũyũ after having built a career writing in English. That is, he analyzed the situation, and then changed his approach.
Maybe the broader artistic community can learn from that. Toward the end of our conversation, Wakatama told me, “I am fervently praying that you’re wrong.” It would be easier if I were wrong; indeed, maybe I am. But if I’m right, it means that several of the world’s most prestigious literary prizes have been infected by language that represents not people’s actual values and cultures, but a damaging mimicry of them by products of the most powerful companies and people in the world. This is the time to do something about it.