The short version
- Zimbabwean author Simbarashe Kundizeza has firmly denied using artificial intelligence to write his Caine Prize-winning short story, The Blue Fighter.
- The controversy stems from an article in The Atlantic that cited Pangram detection software, which flagged the work as entirely AI-generated.
- Critics argue the accusations disproportionately target Black and African writers, while prize officials emphasize the unreliability of untested detection tools.
Simbarashe Kundizeza, the Zimbabwean writer who recently claimed the Caine Prize for African Writing, has issued a direct denial regarding allegations that artificial intelligence composed his winning entry. The controversy erupted after an article in The Atlantic magazine asserted that Pangram, an AI-detection tool, identified Kundizeza’s story, The Blue Fighter, as entirely machine-generated. In a public statement on social media platform X, the author clarified that he wrote the piece himself, describing it as inspired by a poem about fish tanks and reflecting his family’s struggles in Zimbabwe.
The timing of the accusation has drawn sharp criticism from literary leaders. Ellah Wakatama, chair of the Caine Prize board of trustees, expressed shock that The Atlantic published its findings before a thorough investigation could take place. She argued that relying on detection software without corroborating human evidence risks tarnishing both the prize and its winner. Wakatama noted that she was contacted regarding the allegation just days before the article appeared, highlighting the rapid pace at which these claims are circulating in media outlets.
Kundizeza has provided a detailed chronology of his creative process to support his defense. He stated that he possesses a concrete, documented history of the story’s development, including various drafts. As part of an internal review, the Caine Prize is requesting documentation from the author and engaging in ongoing conversations with judges and The Shallow Tales Review, which originally published the work in 2025. Wakatama emphasized that while submissions are not routinely checked for AI, writers vouch for the originality of their work.
The incident has ignited a wider conversation about racial bias in literary scrutiny. Several African writers and publishers have condemned the accusations, suggesting they disproportionately target non-white authors. Wale Lawal, founder of The Republic magazine, described the Atlantic article as unethical and racist. Nigerian publisher Modupe Daramola highlighted the sinister nature of these allegations, questioning why an AI-powered tool like Pangram should be treated as definitive when assessing human creativity.
Joe Ruzvidzo, a Zimbabwean novelist, voiced support for Kundizeza, calling for resistance against what he termed literary stop-and-frisk tactics. The backlash underscores concerns that Black and African writers are facing heightened suspicion in an era where AI detection is becoming more prevalent. Critics argue that these tools are often used as final arbiters despite their known limitations, creating an uneven playing field for authors from marginalized communities.
Vauhini Vara, the author of the Atlantic piece, defended her reporting by citing internal research from Pangram that suggests its assessments are rarely incorrect. She acknowledged her wariness of Silicon Valley technologies but maintained that the tool’s track record warranted attention. Vara also noted that she used her own judgment to evaluate the story, running previous Caine Prize winners through Pangram and finding them to be 100% human-generated. She pointed out a tonal difference between Kundizeza’s winning story and an earlier work from 2021, which the tool identified as human.
However, experts caution against overreliance on such detection methods. Research indicates that AI detectors can frequently misidentify human writing as machine-generated. A study by researchers at the Free University of Brussels found zero errors in Pangram’s assessments of English-language writing by non-native speakers, but it stressed that these tools should only serve as initial flags for further investigation. Wakatama reiterated that Pangram is not yet fully tested and should not be used to judge writers, particularly Black authors, who are often subjected to harsher scrutiny.
The Caine Prize remains committed to determining the truth through proper channels. Wakatama stated she is not opposed to using AI for checking entries but opposes its use as a conclusive measure in this case. The organization continues to gather evidence and consult with stakeholders to ensure a fair resolution. As the debate unfolds, it highlights the growing tension between technological innovation and traditional literary standards.
This episode mirrors similar controversies surrounding other prestigious awards this year, including the Commonwealth Short Story Prize and France’s Goncourt prize. In those instances, winners faced allegations of AI use, leading to withdrawals or intense public debate. The recurring nature of these disputes suggests a systemic issue in how the literary world approaches authenticity in the age of generative AI.
Moving forward, the outcome of the Caine Prize investigation will likely influence how other institutions handle similar allegations. Stakeholders are watching closely to see whether documentation and human judgment can outweigh algorithmic assessments. The case serves as a critical test for the integrity of literary prizes and the fairness with which they treat diverse voices in global literature.
Sources behind this briefing
Go to the original reporting
- BBC World↗Zimbabwean author denies using AI to write award-winning story
