Sok Security Autonomous Llm Agents

---
name: sok-security-autonomous-llm-agents
description: 'Research paper: SoK - Security of Autonomous LLM Agents in Agentic Commerce. Comprehensive security analysis of autonomous agents in commercial settings.'
metadata:
  openclaw:
    emoji: "🔒"
    tags: ["research", "arxiv", "security", "autonomous-agents", "agentic-commerce", "safety"]
---

# SoK: Security of Autonomous LLM Agents in Agentic Commerce

**arXiv ID:** 2604.16227  
**Published:** 2026-04-18  
**Authors:** Qian'ang Mao, Jiaxin Wang, Ya Liu  
**Categories:** cs.CR, cs.MA  
**Utility Score:** 1.0

## Abstract

Autonomous large language model (LLM) agents such as OpenClaw are pushing agentic commerce from human-supervised assistance toward machine actors that can negotiate, purchase services, manage digital assets, and execute transactions across on-chain and off-chain systems. This shift introduces novel security challenges that differ fundamentally from traditional software and human-centric commerce.

## Key Contributions

1. **Agentic Commerce Security**: Systematizes security challenges in autonomous commercial agents
2. **Cross-System Risks**: Analyzes on-chain and off-chain transaction risks
3. **Autonomous Agent Threats**: Identifies unique threats from fully autonomous LLM agents

## Relevance to AI Agent Systems

- **Agentic Systems**: Focus on agentic AI in commercial contexts
- **Autonomous Agents**: Security for fully autonomous LLM agents
- **Tool Use**: Agents using tools for financial transactions
- **Reasoning**: Security implications of agent reasoning

## Technical Keywords

agentic, llm agent, tool, reasoning, benchmark, safety, autonomous

## URL

http://arxiv.org/abs/2604.16227

---

**Tracked:** 2026-04-20  
**Source:** arXiv Paper Tracker