# Breaking the Loop: Detecting and Mitigating Denial-of-Service Vulnerabilities in Large Language Models

arXiv: https://arxiv.org/abs/2503.00416 · HTML: https://mlsanity.com/paper/2503.00416/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 36 references that are in the map.
- First posted on arXiv 1 Mar 2025.

## Builds on (most cited first, 16 of 16)

- [Charformer: Fast Character Transformers via Gradient-based Subword Tokenization](https://mlsanity.com/paper/2106.12672/) (2021)
- [DeepGauge: Multi-Granularity Testing Criteria for Deep Learning Systems](https://mlsanity.com/paper/1803.07519/) (2018)
- [Simulating Classroom Education with LLM-Empowered Agents](https://mlsanity.com/paper/2406.19226/) (2024)
- [Sponge Examples: Energy-Latency Attacks on Neural Networks](https://mlsanity.com/paper/2006.03463/) (2020)
- [FuzzLLM: A Novel and Universal Fuzzing Framework for Proactively Discovering Jailbreak Vulnerabilities in Large Language Models](https://mlsanity.com/paper/2309.05274/) (2023)
- [Energy-Latency Attacks via Sponge Poisoning](https://mlsanity.com/paper/2203.08147/) (2022)
- [Glitch Tokens in Large Language Models: Categorization Taxonomy and Effective Detection](https://mlsanity.com/paper/2404.09894/) (2024)
- [Phantom Sponges: Exploiting Non-Maximum Suppression to Attack Deep Object Detectors](https://mlsanity.com/paper/2205.13618/) (2022)
- [Deconstructing The Ethics of Large Language Models from Long-standing Issues to New-emerging Dilemmas: A Survey](https://mlsanity.com/paper/2406.05392/) (2024)
- [When to Stop? Towards Efficient Code Generation in LLMs with Excess Token Prevention](https://mlsanity.com/paper/2407.20042/) (2024)
- [Automatic Curriculum Learning With Over-repetition Penalty for Dialogue Policy Learning](https://mlsanity.com/paper/2012.14072/) (2020)
- [PerfCE: Performance Debugging on Databases with Chaos Engineering-Enhanced Causality Analysis](https://mlsanity.com/paper/2207.08369/) (2022)
- [Maat: Performance Metric Anomaly Anticipation for Cloud Services with Conditional Diffusion](https://mlsanity.com/paper/2308.07676/) (2023)
- [LLM vs. Lawyers: Identifying a Subset of Summary Judgments in a Large UK Case Law Dataset](https://mlsanity.com/paper/2403.04791/) (2024)
- [Analyzing Diversity in Healthcare LLM Research: A Scientometric Perspective](https://mlsanity.com/paper/2406.13152/) (2024)
- [On-the-fly Improving Performance of Deep Code Models via Input Denoising](https://mlsanity.com/paper/2308.09969/) (2023)

## Built on by (most cited first, 3 of 3)

- [Bridging AI and Software Security: A Comparative Vulnerability Assessment of LLM Agent Deployment Paradigms](https://mlsanity.com/paper/2507.06323/) (2025)
- [RECUR: Resource Exhaustion Attack via Recursive-Entropy Guided Counterfactual Utilization and Reflection](https://mlsanity.com/paper/2602.08214/) (2026)
- [Can a Single Message Paralyze the AI Infrastructure? The Rise of AbO-DDoS Attacks through Targeted Mobius Injection](https://mlsanity.com/paper/2605.11442/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) have significantly advanced text understanding and generation, becoming integral to applications across education, software development, healthcare, entertainment, and legal services. Despite considerable progress in improving model reliability, latency remains under-explored, particularly through recurrent generation, where models repeatedly produce similar or identical outputs, causing increased latency and potential Denial-of-Service (DoS) vulnerabilities. We propose RecurrentGenerator, a black-box evolutionary algorithm that efficiently identifies recurrent generation scenarios in prominent LLMs like LLama-3 and GPT-4o. Additionally, we introduce RecurrentDetector, a lightweight real-time classifier trained on activation patterns, achieving 95.24% accuracy and an F1 score of 0.87 in detecting recurrent loops. Our methods provide practical solutions to mitigate latency-related vulnerabilities, and we publicly share our tools and data to support further research.