# Spectral Guardrails for Agents in the Wild: Detecting Tool Use Hallucinations via Attention Topology

arXiv: https://arxiv.org/abs/2602.08082 · HTML: https://mlsanity.com/paper/2602.08082/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Machine unlearning, one of 12 directions in LLM safety and security.
- Builds on 15 of its 27 references that are in the map.
- First posted on arXiv 8 Feb 2026.

## Placed in

- LLM safety and security / [Machine unlearning](https://mlsanity.com/safety/machine-unlearning/)

## Builds on (most cited first, 15 of 15)

- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs](https://mlsanity.com/paper/2307.16789/) (2023)
- [Gorilla: Large Language Model Connected with Massive APIs](https://mlsanity.com/paper/2305.15334/) (2023)
- [Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation](https://mlsanity.com/paper/2302.09664/) (2023)
- [The Internal State of an LLM Knows When It's Lying](https://mlsanity.com/paper/2304.13734/) (2023)
- [INSIDE: LLMs' Internal States Retain the Power of Hallucination Detection](https://mlsanity.com/paper/2402.03744/) (2024)
- [Hutch++: Optimal Stochastic Trace Estimation](https://mlsanity.com/paper/2010.09649/) (2020)
- [Hallucination Detection in LLMs Using Spectral Features of Attention Maps](https://mlsanity.com/paper/2502.17598/) (2025)
- [A Probabilistic Framework for LLM Hallucination Detection via Belief Tree Propagation](https://mlsanity.com/paper/2406.06950/) (2024)
- [Internal Representations as Indicators of Hallucinations in Agent Tool Selection](https://mlsanity.com/paper/2601.05214/) (2026)
- [Towards Mechanistic Interpretability of Graph Transformers via Attention Graphs](https://mlsanity.com/paper/2502.12352/) (2025)
- [Prompt-Guided Internal States for Hallucination Detection of Large Language Models](https://mlsanity.com/paper/2411.04847/) (2024)
- [A Graph Signal Processing Framework for Hallucination Detection in Large Language Models](https://mlsanity.com/paper/2510.19117/) (2025)
- [LLM Hallucination Detection: HSAD](https://mlsanity.com/paper/2509.23580/) (2025)

## Built on by (most cited first, 2 of 2)

- [Looking Is Not Picking: An Attention-Segment Account of Tool-Selection Failures in LLM Agents](https://mlsanity.com/paper/2606.16364/) (2026)
- [OpenHalDet: A Unified Benchmark for Hallucination Detection across Diverse Generation Scenarios](https://mlsanity.com/paper/2606.06959/) (2026)

## Abstract (arXiv)

Deploying autonomous agents in the wild requires reliable safeguards against tool use failures. We propose a training free guardrail based on spectral analysis of attention topology that complements supervised approaches. On Llama 3.1 8B, our method achieves 97.7% recall with multi-feature detection and 86.1% recall with 81.0% precision for balanced deployment, without requiring any labeled training data. Most remarkably, we discover that single layer spectral features act as near-perfect hallucination detectors: Llama L26 Smoothness achieves 98.2% recall (213/217 hallucinations caught) with a single threshold, and Mistral L3 Entropy achieves 94.7% recall. This suggests hallucination is not merely a wrong token but a thermodynamic state change: the model's attention becomes noise when it errs. Through controlled cross-model evaluation on matched domains (N=1000, T=0.3, same General domain, hallucination rates 20--22%), we reveal the ``Loud Liar'' phenomenon: Llama 3.1 8B's failures are spectrally catastrophic and dramatically easier to detect, while Mistral 7B achieves the best discrimination (AUC 0.900). These findings establish spectral analysis as a principled, efficient framework for agent safety.