New framework OTora tests LLM agents for reasoning-level denial-of-service attacks

Researchers have developed OTora, a novel framework designed to test the resilience of large language model (LLM) agents against a specific type of attack known as Reasoning-Level Denial-of-Service (R-DoS). This attack method aims to degrade an agent's performance by artificially increasing its reasoning depth or tool usage, rather than by causing outright task failure. OTora employs a two-stage process, utilizing adversarial triggers and genetic search to amplify overthinking while maintaining task accuracy, demonstrating significant latency increases on various agent benchmarks. AI

IMPACT This research highlights a new vulnerability in LLM agents, potentially impacting the reliability and efficiency of deployed AI systems.

RANK_REASON The cluster contains a research paper detailing a new framework for testing LLM agent security. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Gaojie Jin
• GPT-OSS-120B
• LLaMA-70B
• LLM Agents
• Reasoning-Level Denial-of-Service

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →