# Gradient Co-occurrence Analysis for Detecting Unsafe Prompts in Large Language Models

arXiv: https://arxiv.org/abs/2502.12411 · HTML: https://mlsanity.com/paper/2502.12411/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 21 references that are in the map.
- First posted on arXiv 18 Feb 2025.

## Builds on (most cited first, 8 of 8)

- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization](https://mlsanity.com/paper/1610.02391/) (2016)
- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails](https://mlsanity.com/paper/2310.10501/) (2023)
- [Not Just a Black Box: Learning Important Features Through Propagating Activation Differences](https://mlsanity.com/paper/1605.01713/) (2016)
- [ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversation](https://mlsanity.com/paper/2310.17389/) (2023)
- [HateBERT: Retraining BERT for Abusive Language Detection in English](https://mlsanity.com/paper/2010.12472/) (2020)
- [Llama Guard 3 Vision: Safeguarding Human-AI Image Understanding Conversations](https://mlsanity.com/paper/2411.10414/) (2024)
- [ShieldLM: Empowering LLMs as Aligned, Customizable and Explainable Safety Detectors](https://mlsanity.com/paper/2402.16444/) (2024)

## Abstract (arXiv)

Unsafe prompts pose significant safety risks to large language models (LLMs). Existing methods for detecting unsafe prompts rely on data-driven fine-tuning to train guardrail models, necessitating significant data and computational resources. In contrast, recent few-shot gradient-based methods emerge, requiring only few safe and unsafe reference prompts. A gradient-based approach identifies unsafe prompts by analyzing consistent patterns of the gradients of safety-critical parameters in LLMs. Although effective, its restriction to directional similarity (cosine similarity) introduces ``directional bias'', limiting its capability to identify unsafe prompts. To overcome this limitation, we introduce GradCoo, a novel gradient co-occurrence analysis method that expands the scope of safety-critical parameter identification to include unsigned gradient similarity, thereby reducing the impact of ``directional bias'' and enhancing the accuracy of unsafe prompt detection. Comprehensive experiments on the widely-used benchmark datasets ToxicChat and XStest demonstrate that our proposed method can achieve state-of-the-art (SOTA) performance compared to existing methods. Moreover, we confirm the generalizability of GradCoo in detecting unsafe prompts across a range of LLM base models with various sizes and origins.