A Harper's Bazaar UK short story contest winner from June 2026 drew fresh scrutiny after screenshots showed the full text of 'Back and Forth' scoring 100 percent AI-generated on Pangram, prompting Nabeel Qureshi to argue that contests must either verify human authorship or stop banning machine assistance.
*Another* apparently AI-generated story wins a literary prize, this time judged by a panel including the novelist Ruth Ozeki.
Literary prizes need to start including Pangram checks in their process, or else change the rules to make AI writing ok. It’s very simple!
Pangram's low false-positive claims rest on aggregate tests, yet one-off flags on contest entries remain unverified by the magazine or judges, leaving room for hybrid writing or clever prompting to complicate any single score.
Without official statements on authorship checks, literary contests risk either quietly accepting AI-aided work or adding new submission hurdles that could chill participation.
Positive users defend the AI-generated story winning Harper's Bazaar literary prize as quality entertainment that shows AI writing has improved, while negative users call it uninspired slop reflecting poor taste by judges.
What's pangram?
Pangram (pangram.com) is an AI content detector that flags text as machine-generated by analyzing statistical patterns from models like ChatGPT, Claude, and Gemini. ¹
It claims 99.98% accuracy with very low false positives, includes plagiarism checks, and is used by schools and publishers—matching its role in the Harper’s Bazaar story.
It happened again.
*Another* apparently AI-generated story wins a literary prize, this time judged by a panel including the novelist Ruth Ozeki.
Literary prizes need to start including Pangram checks in their process, or else change the rules to make AI writing ok. It’s very simple!
AI is generally a weak fiction writer except for one particular kind of fiction (metaphor-rich, staccato sentences, short & plot light, etc.) which it writes excellently. This happens to be a style that can sometimes do quite well in modern literary fiction short story contests.
*Another* apparently AI-generated story wins a literary prize, this time judged by a panel including the novelist Ruth Ozeki.
Literary prizes need to start including Pangram checks in their process, or else change the rules to make AI writing ok. It’s very simple!
Archive link https://archive.is/20260620030105/https://www.harpersbazaar.com/uk/culture/a71615663/short-story-competition-winner-2026/
*Another* apparently AI-generated story wins a literary prize, this time judged by a panel including the novelist Ruth Ozeki.
Literary prizes need to start including Pangram checks in their process, or else change the rules to make AI writing ok. It’s very simple!
@nabeelqu literary judges really need to use pangram or at least spend some time understanding what AI is capable of and what AI writing sounds like
@hrishio Badly wrong
the humans, they yearn for AI art
*Another* apparently AI-generated story wins a literary prize, this time judged by a panel including the novelist Ruth Ozeki.
Literary prizes need to start including Pangram checks in their process, or else change the rules to make AI writing ok. It’s very simple!
@nabeelqu @hrishio How so?
@nabeelqu I wonder if this happens because they are very anti-AI due to politics, and don't use AI, and therefore have no sense of what AI writing is like
Pangram is tuned to basically never flag something as AI-written unless it is sure it is actually AI-written. This means its false positive rate is ~negligibly low.
When people say it is easy to fool, they typically mean that you can get a slightly-modified-but-still-AI-written thing past it. Which is true by design (again, it’s trying to avoid false positives per the first paragraph) but not the case we are concerned with here.
Thus, if it says “100% AI-generated” you can be quite sure it is AI generated. It never does this for human-written text.
@AndrewCurran_ yes its ai generated geometry pale grey the kind that, its ai slop the fact that that wins a prize just tells how bad actualyl the taste in literary cirlces are, i really wish it was otherwise but ai writes really really bad if your iq is not really double digit
@danielrmay No, it’s obvious to anyone who uses GPT or Claude. I included the Pangram screenshot but spotted it organically.
Is @NewYorker next ?
*Another* apparently AI-generated story wins a literary prize, this time judged by a panel including the novelist Ruth Ozeki.
Literary prizes need to start including Pangram checks in their process, or else change the rules to make AI writing ok. It’s very simple!
@nabeelqu @danielrmay It’s not obvious and you’re acting dumb in public but it’s 2026 and you have a lot of followers so you’re probably fine
@hrishio @nabeelqu Meh, I've yet to see it show a false positive.
@akanoego @nabeelqu Does it, or would any other marketing threaten the business model?
@nabeelqu Which model pulled this off?
@nabeelqu And another one where it’s not even necessary to use Pangram (though it’s useful to confirm). Could smell this one from the very first sentence.
@tomharari Yes extremely. I just included the Pangram as an additional data point.
@Hiraweb3 @nabeelqu Agreed but looking at the excerpt, it's not like this author even did something special with AI, it's just generic (although obviously edited) GPT slop.
@brianluidog Yeah I think that’s a decent part of it