Moscow Institute of Physics and Technology (National Research University)

The first meeting of the Global Dialogue can be deemed successful if the following outcomes are achieved: a) the basic principles of AI governance are defined and a corresponding document is adopted; b) working groups on priority topics are established, their terms of reference are defined, and a roadmap for the activities of these working groups is approved.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models

The importance of the thematic block "Secure, Protected and Trusted AI" stems from the critical need to minimise risks of misuse or unqualified use of AI technologies, enhance reliability, build public trust in these technologies, and create conditions for sustainable development - both at the national and global levels. Capacity-building in the field of AI is critically important primarily to reduce global inequality caused by imbalances in resources, infrastructure, and access to technology between developed and developing countries. Furthermore, narrowing the global technology and AI infrastructure gap can mitigate risks arising from the lack of preparedness of developing countries in the AI domain. It is also critically important to prioritise transparency and accountability of AI, as well as human oversight of AI systems, in order to prevent uncontrolled behaviour of AI systems and minimise the associated risks of undesirable consequences - ranging from incorrect medical diagnoses to man-made accidents and disasters. Open-source software, open data, and open AI models also make a substantial contribution to reducing global inequality, as they provide countries and organisations with limited resources access to AI technologies. This enables developing countries and regions to participate in the digital economy, adapt AI to local needs, and close the technology gap.

In our view, the list of topics above is sufficient to launch the work of the Global Dialogue. The list may be expanded in the future as new aspects of AI governance emerge.

In our view, the availability of open‑source software and AI models has the most significant impact on the development of the Russian Federation. It stimulates the growth of high‑tech small businesses that are flexible and quickly adapt to changes, and enhances the effectiveness of collaboration between companies, research organisations, and the government. The availability of open data—including data from emerging government open‑data platforms—also creates conditions for analysis, modelling, and the development of new products and services, as well as for identifying hidden dependencies and eliminating inefficiencies.

The Global Dialogue can play a pivotal role in enhancing international cooperation on AI governance by serving as a unified, inclusive platform for multilateral discussion of best practices and initiatives. It will help foster a common understanding of the risks and opportunities associated with AI, reduce geopolitical disagreements, and create conditions for jointly addressing global challenges.

The Global Dialogue should build upon AI‑related initiatives already developed by the United Nations and its specialized agencies, and also take into account the experience of regional and sectoral organizations—such as BRICS and the International Telecommunication Union (ITU).

Inclusive participation of various stakeholders can make the activities of the Global Dialogue truly comprehensive and effective, taking into account both technological opportunities and social, ethical, and economic aspects. However, to achieve these goals, it is necessary to ensure balanced participation of representatives from governmental and intergovernmental bodies, as well as research organisations and technology companies from both developed and developing countries.

In our view, many developing countries currently lack sufficient influence on the formation of international standards and policies in the field of AI. This is due to limited access to technologies and resources, and, consequently, an underdeveloped expert capacity. The specific challenges faced by these countries — such as the digital divide, dependence on external technologies, and the need to adapt AI to local conditions — often remain on the periphery of global discussions.

In our view, it is important to combine formats aimed at developing in‑depth analysis and expertise with formats designed to address specific tasks. For instance, the expert panel format with participant rotation is well suited to building expertise. It involves discussions with the participation of recognised specialists, in‑depth examination of complex issues, presentation of key research findings, and discussion of long‑term trends and risks. To address specific tasks, short intensive sessions (1–2 days) involving multidisciplinary teams appear to be the most effective. These sessions allow participants to focus on practical outcomes, rapidly test ideas, and receive feedback.

One of the most significant practices in the field of AI governance and ethical risk mitigation is the Code of Ethics in the Field of Artificial Intelligence, adopted on 26 October 2021. This is a self-regulation instrument for the industry, serving as a compendium of general ethical principles and standards of conduct that all actors involved in AI-related activities (developers, clients, users, and other stakeholders) must follow when creating, deploying, and using artificial intelligence technologies. The Code applies to all stages of the AI life cycle that are not governed by legislation or technical regulations, and is intended for systems used exclusively for civilian purposes.