AI text detectors are terrible. And writers are winning awards because of it

Three of five regional winners of the Commonwealth Short Story Prize were accused by readers and writers of using AI-generated text.

The claims relied on style clues and AI detectors, but the organizers said they do not use AI systems in judging and the authors said their stories were original.

The controversy highlights how unreliable AI detectors can be, especially for verifying authorship in literary prizes and publishing.

It also underscores the broader challenge of distinguishing human writing from text produced by models like ChatGPT or Claude.