Seeking Human Security Consensus: A Unified Value Scale for Generative AI Value Safety

TL;DR

This paper introduces a comprehensive, internationally inclusive framework for assessing and improving value safety in generative AI, highlighting current disparities and proposing a pathway toward shared safety standards.

Contribution

It presents the GVS-Scale, a unified value safety framework for generative AI, including risk taxonomy, an incident repository, and an evaluation benchmark, to promote global safety consensus.

Findings

Significant variation in value safety performance across models

Current systems show uneven and fragmented value alignment

Shared safety foundations are essential for progress

Abstract

The rapid development of generative AI has brought value- and ethics-related risks to the forefront, making value safety a critical concern while a unified consensus remains lacking. In this work, we propose an internationally inclusive and resilient unified value framework, the GenAI Value Safety Scale (GVS-Scale): Grounded in a lifecycle-oriented perspective, we develop a taxonomy of GenAI value safety risks and construct the GenAI Value Safety Incident Repository (GVSIR), and further derive the GVS-Scale through grounded theory and operationalize it via the GenAI Value Safety Benchmark (GVS-Bench). Experiments on mainstream text generation models reveal substantial variation in value safety performance across models and value categories, indicating uneven and fragmented value alignment in current systems. Our findings highlight the importance of establishing shared safety foundations through dialogue and advancing technical safety mechanisms beyond reactive constraints toward more flexible approaches. Data and evaluation guidelines are available at https://github.com/acl2026/GVS-Bench. This paper includes examples that may be offensive or harmful.