Dynamic safety cases for frontier AI

TL;DR

This paper introduces a Dynamic Safety Case Management System (DSCMS) for frontier AI that enables continuous, semi-automated updates to safety arguments, ensuring safety claims stay aligned with evolving system states and risks.

Contribution

It adapts methods from autonomous vehicle safety assurance to create a framework for ongoing safety case updates in frontier AI systems.

Findings

Demonstrated DSCMS on a cyber capabilities safety case template

Proposed integration of DSCMS into governance for safety-critical AI

Outlined challenges and future work for continuous safety assurance

Abstract

Frontier artificial intelligence (AI) systems present both benefits and risks to society. Safety cases - structured arguments supported by evidence - are one way to help ensure the safe development and deployment of these systems. Yet the evolving nature of AI capabilities, as well as changes in the operational environment and understanding of risk, necessitates mechanisms for continuously updating these safety cases. Typically, in other sectors, safety cases are produced pre-deployment and do not require frequent updates post-deployment, which can be a manual, costly process. This paper proposes a Dynamic Safety Case Management System (DSCMS) to support both the initial creation of a safety case and its systematic, semi-automated revision over time. Drawing on methods developed in the autonomous vehicles (AV) sector - state-of-the-art Checkable Safety Arguments (CSA) combined with Safety Performance Indicators (SPIs) recommended by UL 4600, a DSCMS helps developers maintain alignment between system safety claims and the latest system state. We demonstrate this approach on a safety case template for offensive cyber capabilities and suggest ways it can be integrated into governance structures for safety-critical decision-making. While the correctness of the initial safety argument remains paramount - particularly for high-severity risks - a DSCMS provides a framework for adapting to new insights and strengthening incident response. We outline challenges and further work towards development and implementation of this approach as part of continuous safety assurance of frontier AI systems.