Search:
Match:
2 results

Prompt-Based DoS Attacks on LLMs: A Black-Box Benchmark

Published:Dec 29, 2025 13:42
1 min read
ArXiv

Analysis

This paper introduces a novel benchmark for evaluating prompt-based denial-of-service (DoS) attacks against large language models (LLMs). It addresses a critical vulnerability of LLMs – over-generation – which can lead to increased latency, cost, and ultimately, a DoS condition. The research is significant because it provides a black-box, query-only evaluation framework, making it more realistic and applicable to real-world attack scenarios. The comparison of two distinct attack strategies (Evolutionary Over-Generation Prompt Search and Reinforcement Learning) offers valuable insights into the effectiveness of different attack approaches. The introduction of metrics like Over-Generation Factor (OGF) provides a standardized way to quantify the impact of these attacks.
Reference

The RL-GOAL attacker achieves higher mean OGF (up to 2.81 +/- 1.38) across victims, demonstrating its effectiveness.

Research#llm📝 BlogAnalyzed: Jan 3, 2026 06:07

Why You Should Stop ChatGPT's Thinking Immediately After a One-Line Question

Published:Nov 30, 2025 23:33
1 min read
Zenn GPT

Analysis

The article explains why triggering the "Thinking" mode in ChatGPT after a single-line question can lead to inefficient processing. It highlights the tendency for unnecessary elaboration and over-generation of examples, especially with short prompts. The core argument revolves around the LLM's structural characteristics, potential for reasoning errors, and weakness in handling sufficient conditions. The article emphasizes the importance of early control to prevent the model from amplifying assumptions and producing irrelevant or overly extensive responses.
Reference

Thinking tends to amplify assumptions.