PASA: A Principled Embedding-Space Watermarking Approach for LLM-Generated Text under Semantic-Invariant Attacks

Documenti analoghi

• Your Semantic-Independent Watermark is Fragile: A Semantic Perturbation Attack against EaaS Watermark
  di: Fei, Zekun, et al.
  Pubblicazione: (2024)
• Blind PRNG Hijacking: An Undetectable Integrity-Preserving Attack Against LLM Watermarking
  di: You, Ziyang, et al.
  Pubblicazione: (2026)
• Modification and Generated-Text Detection: Achieving Dual Detection Capabilities for the Outputs of LLM by Watermark
  di: Cai, Yuhang, et al.
  Pubblicazione: (2025)
• Enhancing LLM Watermark Resilience Against Both Scrubbing and Spoofing Attacks
  di: Shen, Huanming, et al.
  Pubblicazione: (2025)
• PASA: Attack Agnostic Unsupervised Adversarial Detection using Prediction & Attribution Sensitivity Analysis
  di: Bhusal, Dipkamal, et al.
  Pubblicazione: (2024)