Document is current

Article History

Received: 15 April 2025

Accepted: 24 December 2025

First Online: 6 March 2026

Declarations

:

: This study aims to enhance the generative security of T2V models in open environments and to strengthen their reliability in real-world applications. While the research involves the analysis and generation of potentially harmful content, all such content is strictly used for experimental and defense evaluation purposes, in full compliance with academic ethical standards, and is prohibited from any form of dissemination or misuse. We hope that this work will offer theoretical foundations and methodological insights for future research in security assessment, adversarial testing, and defense development, thereby contributing to the advancement of generative models along a compliant and secure trajectory.