AccountabilitySafetyTransparencycs.CY
Blueprints of Trust: AI System Cards for End to End Transparency and Governance
Huzaifa Sidhpurwala, Emily Fox, Garth Mollett, Florencio Cano Gabarda, Roman Zhukov
September 23, 2025
arXiv.org
4 citations
1 influential
Key Takeaways
-
Standardized safety documentation is essential: The paper introduces the Hazard-Aware System Card (HASC) as a unified framework that creates a single, accessible record of an AI system's security and safety information, helping developers and stakeholders make better-informed decisions throughout an AI system's lifecycle.
-
New identification system for AI-specific problems: The framework proposes a novel "AI Safety Hazard (ASH) ID" system that works alongside existing security identifiers like CVEs, enabling clearer and more consistent communication about AI-specific flaws and their fixes—similar to how the tech industry tracks software vulnerabilities.
-
Transparency requires building on existing standards: Rather than starting from scratch, the HASC builds upon established model card and system card concepts, and the paper demonstrates how these AI system cards can complement the ISO/IEC 42001:2023 standard to provide greater overall transparency and accountability.
-
Dynamic safety tracking matters: The framework emphasizes maintaining a "dynamic record" of an AI system's safety posture, suggesting that AI safety documentation should be continuously updated rather than static, reflecting the evolving nature of risks and fixes throughout a system's deployment.
Abstract
This paper introduces the Hazard-Aware System Card (HASC), a novel framework designed to enhance transparency and accountability in the development and deployment of AI systems. The HASC builds upon existing model card and system card concepts by integrating a comprehensive, dynamic record of an AI system's security and safety posture. The framework proposes a standardized system of identifiers, including a novel AI Safety Hazard (ASH) ID, to complement existing security identifiers like CVEs, allowing for clear and consistent communication of fixed flaws. By providing a single, accessible source of truth, the HASC empowers developers and stakeholders to make more informed decisions about AI system safety throughout its lifecycle. Ultimately, we also compare our proposed AI system cards with the ISO/IEC 42001:2023 standard and discuss how they can be used to complement each other, providing greater transparency and accountability for AI systems.