A prize-winning story: AI-written, or not? The detector and the jury differ

A dispute has opened over this year's Caine Prize winner. The detection tool Pangram classifies the winning story as almost entirely AI-generated. The prize's own investigation found "no evidence" of AI use. Both can't be right, and neither side has settled it publicly. The case shows how hard authorship has become to prove. Detectors give probabilities, not proof, and a literary prize has no easy way to demand drafts or process. The next time a story wins a major award, someone will probably ask the same question again.