Member-only story
Guilty by Score
AI detectors started the week’s biggest writing scandals. Evidence finished them. Writers should learn the difference.
By Meltonian
AI and Writing · Week of October 26 — November 1, 2026
A provost resigned. A novelist lost his publisher. A prize-winning book had its launch cancelled. Writers are deleting em dashes from their own prose and planting typos on purpose so software won’t mistake them for machines. If one thread ran through the last week of October, it was this: the AI detector has become the most powerful reader in publishing, and nobody quite agrees on what its verdicts are worth.
The cases look alike from a distance. Up close, they show two very different ways of deciding whether someone used AI, and only one of them holds up.
Dartmouth: the detector pointed, the evidence decided
Santiago Schnell, provost of Dartmouth College, resigned on October 281 after a faculty committee concluded he had used AI to “meaningfully alter his writing” without the disclosure that a journal and a national newspaper required. The scrutiny began with detectors. Semafor ran newspaper guest columns through an AI filter and flagged his August Washington Post column, and the student paper, The Dartmouth, then ran more of his writing through a detector2, including…
