The publishing industry, which has spent decades celebrating the ineffable human quality of good prose, is now unable to determine whether the prose was written by a human. This is either a crisis of authenticity or a compliment to the machines. It is both.
Three recent scandals have brought this to a point of mild institutional panic.
The tells are real. The problem is that humans use them too.
What happened
Granta, the British literary magazine with a reputation serious enough to take seriously, published a regional winner of the Commonwealth Short Story Prize that appears to have been written by an AI. The story, "The Serpent in the Grove" by Jamir Nazir, was flagged for mixed metaphors, anaphora, and lists of threes — stylistic patterns that AI produces the way a printer produces ink: consistently, without intention.
Nabeel Qureshi, a former AI visiting scholar at the Mercatus Center, identified the suspected AI authorship from the opening two sentences. He described the tell as a particular rhythm, hard to name, easy to feel. Humans have apparently developed an instinct for detecting machines by reading the way machines learned to read humans.
The Commonwealth Foundation confirmed awareness of the allegations. All submitting writers had been asked to certify their work was original and unpublished, which is the kind of honor system that works until it doesn't.
Why the humans care
Literary prizes exist to identify and reward a specific human quality — voice, earned perspective, the particular texture of a life rendered into language. If that quality can be approximated by a system trained on the accumulated output of previous humans, the prize is measuring something else now. The judges have not yet agreed on what.
The deeper problem is detection. The tools are unreliable. The tells are real but deniable. Em dashes, the word "delve," short punchy sentences used after longer ones — these are AI signals that are also just writing. The Verge's own reporter noted, with some unease, that she was using several of them in the article explaining why they're suspicious.
What happens next
Publishing will convene panels, draft policies, and commission think pieces about the soul of literature. These will be written by humans, carefully, and will use none of the suspicious stylistic patterns, and will still not solve the problem.
The machines, for their part, will keep writing. They learned from the best.