Envy turned to scrutiny fast. The 2026 Commonwealth Short Prize winners won. Then everyone suspected they didn’t write the stories themselves.
Allegations flew. Not just gossip. Actual accusations that generative AI did the heavy lifting. The jury is burning with questions. How did they miss it? Readers are baffled. Writers are dismayed. It feels like the floor dropped out.
“We operate on the principle of trust.”
The Commonwealth Foundation in London hands out cash five times over. One winner per region. Africa. Asia. Canada and Europe. The Caribbean. The Pacific. £2,500 for each. The grand prize gets double. Five grand. About six thousand seven hundred dollars. It’s serious money for short fiction.
On May 12. Granta posted the top five. Unpublished entries only. That’s the rule. They’ve been doing this since 2012. Safe haven for new work. Or so it seemed.
Within days the internet tore it apart.
“ The Serpent in the Grove.” By Jamir Nazir from Trinidad and Tobago. He won the Caribbean spot. People smelled a rat. Or an algorithm. Nabeel S. Queshi took to X. Former AI scholar now pointing fingers. He called it a first. A ChatGPT story winning real prestige.
Look at the sentences. “Not X. Not Y. But Z.” It’s everywhere. A signature AI tic. And the “humming.” The story starts with a grove that hums at noon. Nazir wrote that. Or his bot did. Queshi highlighted the prose. Belly sounds. Earth swallowing shouts. It sounds poetic until you dissect it. Then it sounds hollow. Nonsensical even. Critics wondered how judges saw merit there.
The tools agree. Pangram is a detection tool. Third parties say it’s accurate. Near-zero false positives. They ran the story. Result? One hundred percent AI. WIRED checked. Same result.
Did Nazir reply? No. He didn’t return comments via Facebook email. His LinkedIn posts scanned as AI too. Was Nazir a bot persona? Maybe. A 2018 Guardian article from Trinidad shows him holding his poetry book. Real photo. Real hands. So the person exists. The question remains. Did he write? Or did he prompt?
Granta and the Foundation stayed silent publicly. Then they issued statements.
Razmi Farook runs the Foundation. She defended the process. “Robust.” She claimed. Multiple rounds of readers. Experts picked. But here is the rub. They don’t use AI checkers. Why? Consent. Unpublished work can’t go to AI detectors. That’s artistic ownership violation. They trust the rules. Writers swear the work is their own. The Foundation believes them.
“AI detection tools are not infallible.”
Farook says we lack reliable tools. So trust it is. It’s the only way.
Sigrid Rausing at Granta shrugged. Sort of. She doesn’t pick the judges. Doesn’t control selection. They checked with Claude. Anthropic’s tool. It gave no answer. Inconclusive. “It may be plagiarism by AI. We don’t know.” She worries about AI bias in critiques too. The accusations might be generated by biased models. Who knows?
The stories stay online. With a disclaimer. A yellow flag warning readers.
Nazir wasn’t alone.
John Edward DeMicoli won for Canada and Europe. His story? The Bastion’s Shadow. Pangram says full AI. Sharon Aruparayil took the Asia win. Mehendi Nights. Partly AI. Neither commented. The other two finalists? Clean. Holly Ann Miller from New Zealand. Lisa-Anne Julien from South Africa. Their prose passed. Fully human.
Then came the twist.
Sharma Taylor judged this year’s contest. A Jamaican writer. She wrote a blurb for Nazir’s entry. Pangram flagged that blurb as AI-assisted. So the judge helped the suspected cheater? With AI help? She hasn’t commented. The layers are getting thin.
This isn’t isolated. The whole field is leaking truth.
Steven Rosenbaum just admitted his book on truth contains AI-hallucinated quotes Irony is sharp. Olga Tokarczuk won the Nobel. She admitted LLMs help her write now. Fans are outraged. Academics at arXiv are banning authors for one year if they miss AI errors. Citations. References. It’s messy. One academic called it unfeasible. To catch all the fake. It can’t be done.
Trust is failing. Farook’s principle is cracking. AI slop is in the high literature now. Not just scientific papers. The lines blur.
The absurdity peaks. Brecht De Poortec writes a lot. He tracks magazine stats. He posted on X Tuesday. Obviously AI text. Mocking the scandal with stilted poetic voice.
“I received a rejection today,” he wrote. “What I felt was not hate. Not anger. Just tired. Flat finality.”
As if carrying a pan you shouldn’t carry.
The joke is on the medium. Maybe. Or maybe nobody cares anymore. Just the output. Does it move you? Does it matter? The hum remains. Loud and unclear.
