On June 30, 2026, Eastern Time, the UN high-level meeting “Building Strategic Capacity: Addressing the Misuse of Artificial Intelligence and Emerging Technologies” was held at United Nations Headquarters in New York. Tang Xing, Chairman of Grouphorse Group, and Guo Lingxin, Director of AI Data and Strategic Partnerships at Grouphorse Group, attended the meeting. During the meeting, UN officials, representatives of various countries, and heads of international organizations engaged in in-depth discussions and dialogue on trends in artificial intelligence, potential risks of misuse, and pathways for systemic governance.

▲ High-Level United Nations Meeting on Artificial Intelligence and Emerging Technologies at UN Headquarters

This session was one of the key topics of the Fourth High-Level Meeting of Heads of United Nations Counter-Terrorism Agencies. It was co-chaired by Abdulaziz Al-Wasil, Permanent Representative of Saudi Arabia to the United Nations, and Akiko Noda, Assistant Secretary-General of the United Nations and Director of the Crisis Bureau of the United Nations Development Programme.

UN Under-Secretary-General and Special Envoy for Digital and Emerging Technologies Amandeep Singh Gill, Mohammed Al-Kuwaiti, Head of the UAE Cybersecurity Council, and Sun Lei, Deputy Permanent Representative of China to the United Nations, among others, attended and participated in the discussions. Representatives from Hungary, Kenya, the Republic of Korea, Spain, Uzbekistan, and relevant UN agencies also shared their views from the perspectives of national security, cyber governance, capacity building, and international cooperation.

Although the meeting was held against the broader backdrop of addressing terrorism and transnational security risks, the issues discussed extended far beyond a single security domain. As artificial intelligence evolves from an information-generation tool into an agent capable of autonomous planning, tool use, task execution, and interaction with real-world systems, the nature of AI risks is also undergoing fundamental change. Governance is no longer concerned only with what a model “says,” but also with how it makes decisions, which tools it invokes, whether it crosses permission boundaries, and whether its behavior can be tested, traced, explained, and promptly halted.

This has direct implications for the rapidly developing AI Agent industry: the core competitiveness of future AI systems will depend not only on model capabilities, but also on whether systems are trustworthy, controllable, and assessable.

▲ Remarks by Amandeep Singh Gill, UN Under-Secretary-General and Special Envoy for Digital and Emerging Technologies

Innovation and Governance Are Becoming a Shared Imperative for AI Development

On June 29, shortly before the UN meeting, the Executive Meeting of the State Council of China made new arrangements for AI development, explicitly calling for stronger high-quality data supply, deeper implementation of the “AI+” initiative, and stronger international cooperation on AI governance.

From China’s high-level policy deployment to multilateral discussions at UN Headquarters on AI risks and capacity building, an increasingly clear development logic is emerging: high-quality data is a critical foundation for AI innovation and large-scale application, while international governance cooperation is an important safeguard for AI to develop in a trustworthy, controllable, and responsible manner. Innovation and governance are not mutually restrictive; rather, they are two capabilities that must be developed in tandem as AI enters high-value, high-responsibility settings such as government affairs, healthcare, finance, public security, and international organizations.

High-Quality Data Is Becoming Infrastructure for Trustworthy AI

For years, Grouphorse has continuously upgraded its business from traditional data processing toward high-quality AI data, human feedback, model evaluation, and intelligent data governance.

In Grouphorse’s view, high-quality data is neither simple data accumulation nor repetitive, low-value annotation. Instead, guided by clearly defined application scenarios, it transforms raw data into data assets that can support model training, evaluation, and governance through rule design, professional judgment, human feedback, quality review, and continuous iteration.

As AI moves beyond content generation toward task execution, the scope and boundaries of high-quality data services are also expanding. In the AI Agent field, such services are no longer limited to supplying data for model training; they also require systematic evaluation sets for complex tasks to test agents’ task understanding, step planning, risk identification, permission compliance, tool use, abnormal termination, and cross-lingual safety performance. This means that data service providers are gradually moving beyond data production into model evaluation, agent testing, and behavioral governance.

▲ Grouphorse Group Chairman Tang Xing and Director of AI Data and Strategic Partnerships Guo Lingxin at the “Global AI Governance Dialogue” sign at the United Nations

Tang Xing noted that the essence of AI governance is by no means to “constrain” innovation, but to lay the foundation for AI to take root in more complex and higher-value business scenarios. Today, AI is rapidly moving from “content generation” to “task execution,” and future industry competition will move beyond competition over parameters, computing power, and model performance alone toward comprehensive assessment of enterprises’ capabilities in systematic evaluation, risk identification, and governance implementation. High-quality data, model evaluation, and behavioral boundaries have become new-generation industrial infrastructure.

Against this backdrop, the value proposition of data enterprises is being reshaped: work with lasting value is not merely about telling a model “what is correct,” but also about teaching it how to act rationally in complex scenarios, when to firmly refuse, and how to promptly mitigate losses when risks emerge.

Connecting International Governance Discussions with Industry Practice

Drawing on multilingual data capabilities, university collaboration networks, experience serving international organizations, and methodologies accumulated in data annotation, quality management, and professional language services, Grouphorse is further exploring high-quality data production, complex-task evaluation, cross-lingual risk identification, and behavioral governance for AI Agents.

In support of international cooperation on AI governance, Grouphorse will further leverage its strengths in cross-lingual collaboration, international project delivery, and global resource connectivity, while actively promoting cooperation with international organizations, technology companies, universities, and research institutions. It will help better connect China’s data capabilities and governance practices with international rules, global application scenarios, and multilateral cooperation needs.

Grouphorse aims to gradually develop service capabilities spanning the entire AI lifecycle:

From scenario definition to data supply; from model training to capability evaluation; and from risk identification to implementation of governance rules.

▲ Grouphorse Group Chairman Tang Xing Attends the UN High-Level Meeting on AI and Emerging Technologies

From China’s accelerated implementation of the “AI+” initiative and active participation in and promotion of international cooperation on AI governance to the UN’s continued multilateral discussions on AI governance, artificial intelligence is entering a new stage of coordinated advancement in innovation, application, and governance.

Grouphorse will continue to use high-quality data as its foundation and model evaluation and behavioral governance as extensions, advancing data capabilities from “supporting model training” toward “supporting trustworthy intelligence.” At the same time, drawing on its multilingual capabilities, experience serving international organizations, and global cooperation network, Grouphorse will help connect China’s data governance practices with international AI governance needs and provide professional support for the large-scale, international, and responsible application of AI.