ML Security PapersDAMAGE: Detecting Adversarially Modified AI Generated Text
Elyas Masrour, Bradley Emi, Max Spero · Pangram Labs
Builds a humanizer-robust AI text detector via data-centric augmentation and validates it against adversarial fine-tuned evasion models
AI humanizers are a new class of online software tools meant to paraphrase and rewrite AI-generated text in a way that allows them to evade AI detection software. We study 19 AI humanizer and paraphrasing tools and qualitatively assess their effects and faithfulness in preserving the meaning of the original text. We show that many existing AI detectors fail to detect humanized text. Finally, we demonstrate a robust model that can detect humanized AI text while maintaining a low false positive rate using a data-centric augmentation approach. We attack our own detector, training our own fine-tuned model optimized against our detector's predictions, and show that our detector's cross-humanizer generalization is sufficient to remain robust to this attack.