Towards Realistic Guarantees: A Probabilistic Certificate for SmoothLLM

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kumarappan, Adarsh, Mehrotra, Ayushi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914377111699456
author Kumarappan, Adarsh
Mehrotra, Ayushi
author_facet Kumarappan, Adarsh
Mehrotra, Ayushi
contents The SmoothLLM defense provides a certification guarantee against jailbreaking attacks, but it relies on a strict "k-unstable" assumption that rarely holds in practice. This strong assumption can limit the trustworthiness of the provided safety certificate. In this work, we address this limitation by introducing a more realistic probabilistic framework, "(k, $\varepsilon$)-unstable," to certify defenses against diverse jailbreaking attacks, from gradient-based (GCG) to semantic (PAIR). We derive a new, data-informed lower bound on SmoothLLM's defense probability by incorporating empirical models of attack success, providing a more trustworthy and practical safety certificate. By introducing the notion of (k, $\varepsilon$)-unstable, our framework provides practitioners with actionable safety guarantees, enabling them to set certification thresholds that better reflect the real-world behavior of LLMs. Ultimately, this work contributes a practical and theoretically-grounded mechanism to make LLMs more resistant to the exploitation of their safety alignments, a critical challenge in secure AI deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2511_18721
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Towards Realistic Guarantees: A Probabilistic Certificate for SmoothLLM
Kumarappan, Adarsh
Mehrotra, Ayushi
Machine Learning
Artificial Intelligence
The SmoothLLM defense provides a certification guarantee against jailbreaking attacks, but it relies on a strict "k-unstable" assumption that rarely holds in practice. This strong assumption can limit the trustworthiness of the provided safety certificate. In this work, we address this limitation by introducing a more realistic probabilistic framework, "(k, $\varepsilon$)-unstable," to certify defenses against diverse jailbreaking attacks, from gradient-based (GCG) to semantic (PAIR). We derive a new, data-informed lower bound on SmoothLLM's defense probability by incorporating empirical models of attack success, providing a more trustworthy and practical safety certificate. By introducing the notion of (k, $\varepsilon$)-unstable, our framework provides practitioners with actionable safety guarantees, enabling them to set certification thresholds that better reflect the real-world behavior of LLMs. Ultimately, this work contributes a practical and theoretically-grounded mechanism to make LLMs more resistant to the exploitation of their safety alignments, a critical challenge in secure AI deployment.
title Towards Realistic Guarantees: A Probabilistic Certificate for SmoothLLM
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.18721