Spider-Sense: Intrinsic Risk Sensing for Efficient Agent Defense with Hierarchical Adaptive Screening

2026-02-05

AI Summary

Current defense mechanisms for large language model agents predominantly rely on mandatory checks, lacking intrinsic awareness and selectivity, thereby struggling to balance security and efficiency. This work proposes Spider-Sense, a novel event-driven defense framework grounded in Intrinsic Risk Sensing (IRS), which activates a tiered response mechanism only upon risk detection. By integrating lightweight similarity matching with internal deep reasoning, Spider-Sense ensures robust security while significantly reducing computational overhead. Notably, the framework operates without external models and enables dynamic trade-offs between efficiency and accuracy. Evaluated on S²Bench—a newly introduced lifecycle-aware benchmark—Spider-Sense achieves the lowest attack success and false positive rates among existing approaches, introducing merely 8.3% latency overhead and demonstrating superior or comparable performance across the board.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Adversarial Agents

Application Category

• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Agentic search
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

As large language models (LLMs) evolve into autonomous agents, their real-world applicability has expanded significantly, accompanied by new security challenges. Most existing agent defense mechanisms adopt a mandatory checking paradigm, in which security validation is forcibly triggered at predefined stages of the agent lifecycle. In this work, we argue that effective agent security should be intrinsic and selective rather than architecturally decoupled and mandatory. We propose Spider-Sense framework, an event-driven defense framework based on Intrinsic Risk Sensing (IRS), which allows agents to maintain latent vigilance and trigger defenses only upon risk perception. Once triggered, the Spider-Sense invokes a hierarchical defence mechanism that trades off efficiency and precision: it resolves known patterns via lightweight similarity matching while escalating ambiguous cases to deep internal reasoning, thereby eliminating reliance on external models. To facilitate rigorous evaluation, we introduce S$^2$Bench, a lifecycle-aware benchmark featuring realistic tool execution and multi-stage attacks. Extensive experiments demonstrate that Spider-Sense achieves competitive or superior defense performance, attaining the lowest Attack Success Rate (ASR) and False Positive Rate (FPR), with only a marginal latency overhead of 8.3\%.

Problem

• agent defense
• intrinsic risk sensing
• security validation
• autonomous agents
• large language models

Innovation

• Intrinsic Risk Sensing
• Hierarchical Adaptive Screening
• Event-driven Defense
• LLM Agent Security
• S²Bench