Agentic Adversarial Rewriting Exposes Architectural Vulnerabilities in Black-Box NLP Pipelines

Documenti analoghi

• Image Safeguarding: Reasoning with Conditional Vision Language Model and Obfuscating Unsafe Content Counterfactually
  di: Bethany, Mazal, et al.
  Pubblicazione: (2024)
• CAMOUFLAGE: Exploiting Misinformation Detection Systems Through LLM-driven Adversarial Claim Transformation
  di: Bethany, Mazal, et al.
  Pubblicazione: (2025)
• Can Reinforcement Learning Unlock the Hidden Dangers in Aligned Large Language Models?
  di: Karkevandi, Mohammad Bahrami, et al.
  Pubblicazione: (2024)
• Deciphering Textual Authenticity: A Generalized Strategy through the Lens of Large Language Semantics for Detecting Human vs. Machine-Generated Text
  di: Bethany, Mazal, et al.
  Pubblicazione: (2024)
• Improving LLM Reasoning with Multi-Agent Tree-of-Thought Validator Agent
  di: Haji, Fatemeh, et al.
  Pubblicazione: (2024)