Poynter: AI-writing scandals are common now. Proving them is harder.

In May 2026, allegations of AI-generated writing hit two high-profile literary publications in quick succession: the Commonwealth Short Story Prize and Granta. Angela Fu’s Poynter commentary looks squarely at what these cases share — not that accusations were made, but that no one could prove anything either way.

The detection problem is technical and not easily solvable in the near term. Purpose-built AI detectors produce false positives and false negatives at rates that undermine their usefulness as evidence. General-purpose models perform worse still. As Atlantic contributor Vauhini Vara notes, AI output trained to mimic human voice can pass tools designed to catch it, particularly when writers have used AI as a drafting assistant rather than a pure ghostwriter. The distinction between “AI-assisted” and “AI-generated” is invisible to current software and arguably meaningless at certain points in the spectrum.

This creates an asymmetry of professional consequences. Writers with established reputations have strong incentive to defend themselves publicly and in detail. Writers who are newer or less prominent have less to gain by engaging, and accusations can sit unresolved. Fu argues that without technical proof, public shaming functions as the primary enforcement mechanism — and that accused writers are becoming aware they can sit out the cycle without facing formal consequences.

For editors and publications writing AI policies, the piece offers a useful corrective to overconfidence in detection tooling. The practical governance question is shifting from “how do we catch AI writing after submission” to “how do we design submission and editorial processes that make substitution less likely in the first place.” That is a structural problem, not a technological one, and it puts the responsibility back on commissioning editors and the relationships they maintain with contributors. The absence of reliable detection makes those relationships the main quality control mechanism.

Useful for editors building AI submission policies, literary magazine staff facing similar controversies, and writers thinking about how to document their process as a form of professional protection.