LATIDIA · Ciberseguridad

StealthRL: Ataques de paráfrasis de aprendizaje de refuerzo para la evasión multidetector de detectores de texto de IA

arXiv:2602.08934v3 Anuncio Tipo: replace-cross Resumen: Los detectores de texto basados ​​en IA se utilizan cada vez más en entornos de alto riesgo, pero su robustez frente a la reescritura adversaria que preserva el significado sigue sin estar clara. Presentamos St

La noticia

arXiv:2602.08934v3 Announce Type: replace-cross Abstract: AI-text detectors are increasingly used in high-stakes settings, but their robustness to meaning-preserving adversarial rewriting remains unclear. We introduce StealthRL, a reinforcement learning framework for stress-testing detectors with adaptive paraphrase attacks. StealthRL trains a paraphrase policy against a detector ensemble while enforcing semantic fidelity, then evaluates transfer to held-out detector families on the full filtered MAGE test pool. Across RoBERTa, Fast-DetectGPT, Binoculars, and MAGE, the resulting attack sharply degrades detection performance, driving mean TPR@1%FPR down to 0.024 while retaining nontrivial semantic overlap and readable prose. The attack also transfers to detectors not used during training, indicating a broader robustness failure rather than overfitting to a