Developing Assurance Cases for Adversarial Robustness and Regulatory Compliance in LLMs

TL;DR

This paper introduces a layered assurance framework for enhancing adversarial robustness and regulatory compliance in large language models, addressing vulnerabilities and dynamic risks across deployment stages.

Contribution

It proposes a novel layered assurance approach with a meta-layer for risk management, tailored to LLM vulnerabilities and compliance requirements.

Findings

Layered framework effectively mitigates adversarial attacks.

Meta-layer enables dynamic risk assessment.

Assurance cases demonstrate tailored strategies for robustness and compliance.

Abstract

This paper presents an approach to developing assurance cases for adversarial robustness and regulatory compliance in large language models (LLMs). Focusing on both natural and code language tasks, we explore the vulnerabilities these models face, including adversarial attacks based on jailbreaking, heuristics, and randomization. We propose a layered framework incorporating guardrails at various stages of LLM deployment, aimed at mitigating these attacks and ensuring compliance with the EU AI Act. Our approach includes a meta-layer for dynamic risk management and reasoning, crucial for addressing the evolving nature of LLM vulnerabilities. We illustrate our method with two exemplary assurance cases, highlighting how different contexts demand tailored strategies to ensure robust and compliant AI systems.