When Efficiency Backfires: Cascading LLMs Trigger Cascade Failure under Adversarial Attack

Documenti analoghi

• Cascade: Composing Software-Hardware Attack Gadgets for Adversarial Threat Amplification in Compound AI Systems
  di: Banerjee, Sarbartha, et al.
  Pubblicazione: (2026)
• When Backdoors Go Beyond Triggers: Semantic Drift in Diffusion Models Under Encoder Attacks
  di: Chen, Shenyang, et al.
  Pubblicazione: (2026)
• CAVGAN: Unifying Jailbreak and Defense of LLMs via Generative Adversarial Attacks on their Internal Representations
  di: Li, Xiaohu, et al.
  Pubblicazione: (2025)
• When Grammar Guides the Attack: Uncovering Control-Plane Vulnerabilities in LLMs with Structured Output
  di: Zhang, Shuoming, et al.
  Pubblicazione: (2025)
• Stealthy Backdoor Attacks against LLMs Based on Natural Style Triggers
  di: Wei, Jiali, et al.
  Pubblicazione: (2026)