The Ethics of Accusation

Week 11

Wrong 5 to 20% of the time. Treated as a verdict.

This already happened, repeatedly

  • a Johns Hopkins professor had to hand-review a student’s drafts after a detector flagged over 90% of her own paper as machine-written
  • two University of California, Davis students were flagged, and separately cleared on appeal
  • in early 2026, a court ordered an Adelphi University student’s “100% AI” finding expunged, calling it without valid basis

The institutional response

More than two dozen universities — MIT, Yale, NYU, UC Berkeley, University of Toronto, UBC, Macquarie, Manchester among them — have restricted or dropped AI-detection tools outright. Not because detecting AI text doesn’t matter. Because a tool wrong one time in five was being handed the authority of a verdict.

A forensic practice that only worries about missing the guilty and never about convicting the innocent is a witch-hunt with better instruments.

Borrowed from actual criminal procedure

  • presumption of innocence — the artefact is not guilty until the evidence says so
  • burden of proof — suspicion is not the same as confidence
  • what you owe someone when you’re wrong — this is not optional just because the accused is a student, not a defendant

Further reading

washingtonpost.com/technology/2023/08/14/prove-false-positive-ai-detection-turnitin-gptzero/

Washington Post — proving your innocence after a false positive