Global Initiative on the Development and Governance of Trustworthy AI Training Data

LONDON, June 15 — As a key official component of London Tech Week 2026, the World Future Technology Development Summit was held on June 11 at IET London: Savoy Place, the Royal engineering venue on the banks of the River Thames. At the summit, the Global Initiative on the Development and Governance of Trustworthy AI Training Data was officially launched. The International Artificial Intelligence Association (IAIA), as a co-initiating organization, attended the launch and contributed to multilateral collaboration on global AI governance.

This year's summit focused on the strategic connection between future technology and international capital, bringing together more than 400 international experts, industry leaders, and representatives of innovation institutions from fields including artificial intelligence, robotics, intelligent manufacturing, sustainable-development technology, policy and governance, and industrial investment, to jointly assess emerging trends and new paradigms in the evolution of global technology. The Initiative was jointly initiated and co-signed by institutions including the Centre for Artificial Intelligence at University College London (UCL), the World Future Technology Development Summit, the International Artificial Intelligence Association, the National Advertising Research Institute, and the UK edition of Nouvelles d'Europe (Europe Times), with academic support from the team led by Professor Song Kai of Beijing Language and Culture University, reflecting a cross-disciplinary, cross-national, and cross-industry consensus on collaborative governance.

I. Trustworthy Training Data Emerges as a New Variable in AI Competition and Governance

The launch noted that the global landscape of AI competition is undergoing a historic turning point: the center of gravity has accelerated from being “algorithm-driven” and “compute-driven” toward a new, “data-driven” stage. As large-model architectures mature, the core variable determining the boundaries of AI capability is shifting from sheer parameter scale and training compute toward a system of trustworthy training data that is high-quality, verifiable, licensable, and traceable. Particularly amid the large-scale deployment of generative AI, systemic risks—the proliferation of misinformation, ambiguous data-copyright ownership, the reverse contamination of training sets by AI-generated content, and insufficient supply for low-resource languages—are becoming critical bottlenecks constraining the trustworthy development of AI worldwide.

As Chairman of the International Artificial Intelligence Association, a co-initiating organization of the Initiative, Amb. Dr. Rui Dai noted that whoever defines what data is authentic and trustworthy will, to a considerable degree, define how AI understands the world—and how it understands human beings themselves. In his view, the governance of trustworthy training data concerns the very foundations of digital civilization, and whether different civilizations can be fairly represented in the intelligent age.

II. Six Standards to Build Trustworthy Training-Data Infrastructure

In response to these challenges, the Initiative proposes building trustworthy training-data infrastructure oriented toward global AI development, and establishes six core standards: verifiable Provenance, lawful and complete Licensing, Factuality, Ethics & Bias Mitigation, Standardization, and Dynamic Updating. This framework aims to systematically reduce AI hallucination, model bias, and compliance risk at the data source, laying a more solid data foundation for high-risk industries such as finance and healthcare, for specialized-knowledge scenarios, and for cross-cultural intelligent communication.

Notably, the launch highlighted UCL's academic leadership in AI governance, AI policy research, and cross-disciplinary technological collaboration. The UCL team has long worked deeply in AI regulation, model safety, social impact, and international cooperation, providing profound academic grounding and a global perspective for the construction of trustworthy training-data standards. By closing the two-way loop between “technological capability” and “governance capability,” the Initiative fosters closer coordination among academic research, industry needs, and policy design.

III. From a Statement of Principles to a Compliance-Engineering Platform

At the level of technical implementation, the summit concurrently launched a global responsible-AI compliance-engineering platform—FluxFormAI—jointly developed by relevant teams at UCL and the University of Cambridge. Centered on core elements such as data compliance, regulatory mapping, safety benchmarking, quality assessment, dynamic iteration, risk grading, and security review, the platform seeks to translate fragmented global AI-governance standards into executable, verifiable, and deployable one-stop engineering solutions, substantively enabling the global rollout of AI technologies and products.

The launch of this platform signals that the building of trustworthy AI no longer remains at the level of principled advocacy, but has entered a new cycle of “engineering the standards and turning governance into infrastructure.” For technology enterprises targeting overseas markets, trustworthy training data and compliance-engineering capability together form key pillars of international competitiveness: the former determines the quality boundary of how a model understands the world, while the latter determines a product's sustainable ability to enter different markets and industry scenarios.

IV. Global Action Plan and Outlook for Cooperation

In addition, the Initiative sets out four key action plans: conducting global training-data audits and ratings, advancing mutual recognition of international standards, building an open and shared platform, and establishing a Global Youth Research Fellowship to cultivate the next generation of AI data-governance talent. Through these measures, the Initiative seeks to connect universities, research institutions, AI enterprises, policymakers, and content creators in jointly building a transparent, fair, compliant, and sustainable AI data ecosystem. As a co-initiating organization, the International Artificial Intelligence Association said it will actively participate in advancing these actions and, drawing on its own international network, contribute to areas such as the mutual recognition of standards and the cultivation of young talent.

The World Future Technology Development Summit provided a strategically significant arena for international exchange around the Initiative. Closely tracking the pulse of UK–China and global technology cooperation, the summit emphasized the deep coupling of frontier research, industrial application, and international capital. Launching the global initiative on trustworthy AI training data on this platform not only establishes the foundational role of AI governance in future technological development, but also underscores the irreplaceable importance of international cooperation in co-building standards and governing technology.

Looking ahead, trustworthy training data will become a strategic infrastructure of the AI era. As the Initiative emphasizes: “Algorithms determine efficiency, computing power determines speed, and trustworthy training data determines the future.” In this critical window, as the rules of global AI competition and governance are rapidly being reshaped, building an open, compliant, traceable, and sustainable system of trustworthy training data will lay the data foundation for responsible AI and provide new public infrastructure for global technology cooperation and the building of digital civilization.

---