WDTA Unveils World’s First AI Agent Security Standard at UN Geneva Event

2025-07-12 18:55:08

GENEVA, July 11 – At a “Global Consultation on the Social Aspects of Digital Technologies and AI” jointly hosted by the United Nations Research Institute for Social Development (UNRISD) and the World Digital Technology Academy (WDTA) at UN European Headquarters in Geneva, WDTA unveiled its new AI STR Series: Single AI Agent Runtime Security Testing Standards.

Speaking at the event, vice-chair of UN Commission on Science and Technology for Development (CSTD) and Honorary Chairman of WDTA Peter Major said: “Ensuring fairness in data governance, and integrating AI safety, ethics, and social values, are key to advancing global sustainable development. To effectively navigate the future of digital technologies, it is essential to establish sound legal systems, collaborative governance frameworks, and comprehensive technical standards—so that technological equity becomes a driving force for sustainable development.”

Prof.Yale Li, Executive Chairman of WDTA, noted that AI agents, capable of deep reasoning, autonomous planning, decision-making, and execution, are now being widely adopted across various domains, including content creation, knowledge services, and business process optimization. However, this technological leap also brings significant safety concerns.

According to a recent report by 21st Century Business Herald, over 70% of professionals working with AI agents express concern about issues such as AI hallucinations, flawed decision-making, and data leakage, as AI-generated content often contains factual inaccuracies or misinterprets user intent. More than half of the respondents also indicated that their organizations have not yet appointed a dedicated officer responsible for AI agent safety.

To address these concerns, Prof. Li introduced the release of WDTA’s new "AI Agent Operational Safety Testing Standards", which provides a systematic, actionable testing framework and evaluation methodology. These standards are designed to be applied across multiple high-risk sectors, including autonomous driving, healthcare, manufacturing, and finance.

This newly issued standard marks the latest milestone under WDTA’s flagship initiative: the AI STR Framework, which promotes the development and deployment of Safe, Trustworthy, and Responsible Artificial Intelligence.