Artificial Intelligence (AI) is undoubtedly a child of the information revolution. This revolution started with the first computers in the 1940s. Communication satellites followed in the 1960s. Since the 1970s, the Internet has become the main infrastructure of the information age. The Internet triggered the development of the digital economy and social networks. And all this paved the way for today´s AI Hype.

Like all technological revolutions, it needed some time to recognize and understand the social and political implications. The second industrial revolution, which began at the end of the 19th century, not only created new opportunities, but it also sparked political conflicts, including two world wars. Insofar as we should not be surprised to see similar ruptures in today´s political landscape. The information revolution and AI are changing everything: from our way of life to the nature of war.

The new opportunities will also lead to new conflicts within and between countries. Are we prepared to handle these new challenges? Do we have the institutional frameworks to deal with these new conflicts? And do we have the political leaders who understand the nature and the risks of this unstoppable information revolution?

The UN as the most universal political platform

AI is a universal problem. With all its weaknesses, the United Nations is the most universal political body we have. In other words, the UN would be the natural venue for discussing global policies and addressing AI challenges.

Dealing with questions arising from the information revolution is not new to the UN. But the history of “UN Information Summits” shows mixed results. There are failures and successes.

Already during the 1st UN General Assembly in 1946, information issues, and here in particular the right to freedom of expression, were the subject of negotiations. UN resolution 59(I) defined freedom of information as “a fundamental human right” and “the touchstone of all the freedoms”. The resolution decided to convene the “UN Conference and Freedom of Information”, which took place in spring 1948 in Geneva. It failed. The shadow of the early days of the Cold War blocked any universal agreement.

In the 1970s and 1980s, both the UN and UNESCO restarted discussions on information issues, triggered by the emergence of communication satellites. In particular, developing countries wanted to have a “New World Information and Communication Order” (NWICO). Also, this project failed. It ended up in a conflict of media control vs. free flow of information and became the victim of the final days of the Cold War. The US left UNESCO in 1983. The Soviet Union collapsed in 1990. And the NWICO books were closed on May 3, 1991, during a UNESCO Conference in Windhoek. Since 1993, May 3 has been celebrated as World Press Freedom Day.

When the Internet arrived on the political agenda in the 1990s, the global political environment was different. There was a growing consensus that globalization will end traditional zero-sum games between governments and offer win-win opportunities, with a broad participation of non-governmental stakeholders in bottom-up policy development processes. In 1998, ICANN did pioneer this new “multistakeholder model”.

This “new thinking” was also behind the UN World Summit on the Information Society (WSIS), which started in 2002. WSIS produced a remarkable consensus. 193 Heads of States declared in December 2003 in Geneva the “common desire and commitment to build a people-centred, inclusive and development-oriented Information Society, where everyone can create, access, utilize and share information and knowledge, enabling individuals, communities and peoples to achieve their full potential in promoting their sustainable development and improving their quality of life, premised on the purposes and principles of the Charter of the United Nations and respecting fully and upholding the Universal Declaration of Human Rights”.

But since 2005, when the 2nd phase of WSIS, in its Tunis Agenda, agreed on the multistakeholder approach to the governance of Internet-related problems, the world has changed again. Zero-Sum Games are back in political battles between big powers.

However, the WSIS consensus, including the multistakeholder definition for Internet Governance, was reiterated just recently during the WSIS+20 Review conference in UN Resolution 80/173, adopted on December 15, 2025.

This should be good news for the beginning of a new round of global discussion within the UN system, on how to manage the consequences of the information revolution, this time related to the development of AI. But the new WSIS consensus is as fragile as today´s political landscape. More than a dozen states expressed their dissatisfaction with the adopted text of the UN Resolution 80/173.

The AI Panel and the Global Dialogue on AI Governance

In this new, fragile political environment, the global discussion around the governance of artificial intelligence has just started. But it does not have to start from Zero. There is no need to reinvent the wheel. There is already a body of norms and principles relevant to AI Governance.

AI Governance is certainly related to Internet Governance. Without the Internet, there would be no AI. AI was on the agenda of the Internet Governance Forum (IGF) since the mid-2010s. In 2021, a “Policy Network for Artificial Intelligence (IGF.PNAI) was formally established as a subsidiary body of the IGF. PNAI discussed how the basic principles for Internet Governance, such as the multistakeholder and the holistic approaches, openness, inclusion, transparency, accountability, bottom-up policy development, and others, can be creatively adopted as a basis for the governance of AI.

Other discussions on AI Governance took place within ITU, UNESCO and OECD. In 2017, the ITU organized its first “AI for Good” Summit. 2019, the OECD adopted five basic principles for AI Governance, which were later supported by the G20. UNESCO adopted in 2021 a “Recommendation on AI and Ethics”. The Council of Europe started work on a legally binding instrument in 2019 which led to the adoption of a “Framework Convention on Artificial Intelligence and human rights, democracy and the rule of law” in September 2024. The EU was pioneering the field by its “AI Act” in 2025.

In 2023, the UN established a “High-Level Advisory Body on Artificial Intelligence”. The recommendations of this group were integrated into the Global Digital Compact (GDC), adopted by the 79th UN General Assembly (UNGA) in September 2024. The GDC decided to create an “Independent International Scientific Panel on AI” and to launch a “Global Dialogue on AI Governance”.

The independent panel, 40 experts from all over the world, was formally established in February 2026. It is co-chaired by Maria Ressa, Peace Nobel Prize Winner and former Co-Chair of the IGF Leadership Panel, and Yoshua Bengio from the University of Montreal. Its first F2F meeting took place in Madrid in April 2026. UN Secretary General Guterres welcomed the panel members with warm words: “No country, no company, and no field of research can see the full picture alone because today AI is reshaping economies and societies. As an engineer, I share your conviction that science and facts matter. And as a longtime politician and diplomat, I have seen how quickly fear can take hold when facts are missing or distorted—how trust breaks down, and division deepens. The world urgently needs a shared, global understanding of artificial intelligence. Grounded not in ideology, but in science, not in fake news, but in knowledge. That is where you come in. Your role is to bring independent, credible science into the global conversation—and to do so at a time when geopolitical tensions are rising, conflicts are raging, and the stakes for safe and responsible AI could not be greater. In this fractured context, an unbiased and trusted understanding of AI is essential.”

The 1st Global Dialogue on AI Governance will take place in July 2026 in Geneva, where they will discuss, inter alia, the first report of the AI Panel. Preparations for the Global Dialogue are underway. A series of stakeholder consultations has started in April and May. The consultations will continue in June 2026. The interest in participating and contributing is overwhelming. Only for the first round of virtual stakeholder consultations, there were 368 requests to speak. More than 1500 written contributions were sent to the UN secretariat.

As it looks now, this could become a success story. But there are some “buts”.

Opposition by the US and alternative Processes

One big “but” is the opposition of the US government. Another “but” is the existence of competing platforms.

The AI Panel is in troubled waters after the US government expressed its opposition. On February 3, 2026, the US voted against the UN Resolution 80/610, which nominated the 40 members of the new AI Panel. During the AI Impact Summit in New Delhi (February 20, 2026), Michael Kraitsos, Director of the White House Office of Science and Technology Policy, was directly attacking both the UN Panel and the UN Dialogue: “As the Trump Administration has now said many times: We totally reject global governance of AI. We believe AI adoption cannot lead to a brighter future if it is subject to bureaucracies and centralized control.”

It has to be seen how this will affect the global discussion. The AI Panel includes two members from the US: Martha Palme from the University of Boulder in Colorado and Vipin Kumar from the University of Minnesota. US AI companies will participate in the global dialogue on AI Governance in July 2026 in Geneva.

The other “But” is related to alternative processes. On the one hand, ITU, with its traditional “AI for Good” Summit, and UNESCO are very supportive. Other bodies, such as the OECD, which has its own “Global Partnership for AI” (GPAI), are more in a “wait and see” position. It is also unclear how the so-called “Bletchley Process” will be linked to the Global Dialogue on AI Governance.

The former British Prime Minister Sunak organized the first global AI Summit in February 2023 in Bletchley, the place where British codebreakers cracked the “Enigma Code” to end World War II. Political leaders—US Vice President Harris, French President Macron, German Chancellor Scholz and others - participated together with CEOs from US and Chinese AI companies. At the end of the meeting, they adopted the “Bletchley Declaration.”

The declaration opens with the view that AI “presents enormous global opportunities: it has the potential to transform and enhance human wellbeing, and prosperity.” With that rosy-tinted view firmly in place, the statement goes on to state the goal: “To realize this, we affirm that, for the good of all, AI should be designed, developed, deployed, and used, in a manner that is safe, in such a way as to be human-centric, trustworthy and responsible.”

Follow-up summits took place in Seoul (2024), Paris (2025) and New Delhi (2026). The summit process will continue in 2027 (Geneva) and 2028 (Abu Dhabi). It remains to be seen how constructive or competitive a relationship will emerge between the UN Dialogue and the Bletchley process.

Another problem is the Chinese proposal for a new “World Artificial Intelligence Cooperation Organization” (WAICO), planned to be headquartered in Shanghai. China supports the UN process. But since July 2025, when China published its global AI Initiative, it has also been advertising on the highest level its WAICO proposal: from President Xi to Prime Minister Li to Foreign Minister Wang. During the recent summit meeting between Russia and China, the Russian Präsident Putin supported this idea. And also, the BRICS Foreign Ministers Meeting in India in May 2026 took note of this idea. How will a WAICO be linked to the UN Panel and the Global Dialogue on AI Governance?

So lets wait and see how the Global Dialogue on AI Governance will find its way into the future discussion landscape between the IGF, Bletchley and WAICO. The 2nd Global Dialogue on AI Governance is already scheduled for June 2027 in New York as part of the annual STI-Forum. The Global Digital Compact (GDC), which created the two new UN AI units, will be reviewed in fall 2027 by the 82nd UN General Assembly.

And just recently, Pope Leo XIV. offered in his Encyclica “Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence” food for thought for the UN Panel and the broader community. The Pope relates his wisdom in many of the 245 paragraphs directly to issues under discussion in the UN, including the negotiations on autonomous weapon systems (GGE LAWS): “The decision to use lethal force cannot be delegated to opaque or automated processes, but must remain under effective, self-aware and responsible human control”, argues the pope and sets a standard, which will hopefully be heard not only by the members of the AI Panel, but also by the negotiators of the next round of negotiations of the GGE LAWS, scheduled for early September 2026 in Geneva.

Interesting times we live in. Stay tuned.

Below is my statement I sent in written form to the UN Secretariat.

Informal Stakeholder Consultations on the Global Dialogue on AI Governance, April, 23, 2026

Thanks, Co-Facilitators, for the arrangement of informal stakeholder consultations for the forthcoming “Global Dialogue on AI Governance”. I have two points. One on substance and one procedures.

What is AI Governance?

There is no consensus what “AI Governance” means. We do not have an agreed definition. Different players are using the AI Governance language with a different understanding. This can lead to confusion. With other words, it would make sense to work towards an “AI Governance Definition”.

More than 20 years ago we did have a similar situation. The world discussed “Internet Governance”, but nobody had a clear understanding what does it mean. The 1st phase of the UN World Summit on the Information Society (WSIS) agreed to establish a multistakeholder “Working Group on Internet Governance” (WGIG) with a mandate, inter alia, to define Internet Governance.

I was one of the 40 members of this group and we did have extensive discussions on a “broad” and a “narrow definition”. Finally, we decided to propose a broad definition, which would go beyond the management of critical Internet resources (CIRs) as domain names, IP addresses and Internet Protocols. We included also Internet related public policy issues as cybersecurity, digital economy and human rights in our definition.

The definition had three elements.

• First: It proposed the multistakeholder approach for the governance of the Internet;
• Second: It supported the philosophy of “sharing”, which was in the DNA of the Internet from its very first day, for “principles, norms, rules, decision-making procedures and programmes”;
• And third: We made clear, that a “holistic approach” is needed, which would encompass both the technical and the political aspects of the Internet.

This Internet Governance definition was finally adopted by the 193 UN member states during the 2nd WSIS phase in paragraph 34 of the “Tunis Agenda”. Just recently, during the WSIS+20 negotiations in New York, this definition was reiterated in the “WSIS+20 Outcome Document” by the 80th UN General Assembly (UNGA) on December 16, 2025.

Since 2005 and 2026, many things have changed in the digital world. However, basic issues and challenges are not so different. We have seen an enormous broadening of the debate with a lot of new governance challenges. Every new digital problem produced the question “How to govern it.” And so we did have discussions on Cyber Governance, IOT Governance, ICT Governance, Digital Governance, Data Governance and now AI Governance.

How all this governance challenges are inter-related? During the NetMundial+10 Conference (April 2024 in Sao Paulo), there was an interesting discussion, how to avoid fragmentation and confusion in this debates. The Sao Paulo outcome document used more general language as “digital policy processes” and “governance of the digital world”.

In my eyes, this could be very relevant for a serious debate about AI Governance. The debate on AI Governance doesn’t start from scratch. Certainly, there are differences between Internet Governance and AI Governance. But there are also similarities. There is no need to reinvent the wheel. The wheel has to be enhanced and improved.

The three elements of the Internet Governance definition from 2005, as mentioned above, are all relevant for AI Governance:

• AI Governance has to be multistakeholder. No stakeholder group alone, neither governments nor the private sector, nor the technical community, nor civil society alone can manage the future of AI. All stakeholders have to work hand in hand to find compromises on common solutions, which takes into consideration legitimate interests of the involved and affected stakeholders. If one stakeholder group tries to dictate what AI governance is, this could lead to subordination, exploitation and political tyranny;
• Ai Governance should also be guided by the philosophy of “sharing”. It has to be open. Open Source and open data are important elements. Transparency, accountability and inclusion are needed as well as bottom-up policy development;
• And AI governance needs a holistic approach. It has to take into consideration both the technical aspects as well as the political, legal, economic, cultural, ethical and social implications.

But there are also differences between Internet Governance and AI Governance. I see two main issues, which are different:

First: AI governance needs a risk-based approach. 25 years ago, the Internet was seen primarily as an enabling technology with incredible opportunities. Risks were mainly related to the technical security of the networks. Today, AI offers many more opportunities, but the risks are higher. The risks are now political. As Yoval Noah Harari said in Davos at the World Economic Forum (WEF) in January this year, AI is not just a tool like a knife, which can be used by human beings for different—good or bad—purposes. AI is a knife, said Harari, which can produce other knives which will go out of human control. Those risks have to be taken into consideration. This has to be a central part of any AI governance. And this leads to the need of regulation. Nobel Prize winner Geoffrey Hinton recently compared AI regulation with rules for car making: “If you ever went out with a car that had no brake, boy, you are in trouble if you go down a hill. But you’re in even more trouble if there’s no steering wheel.” In the car industry, brakes and steering wheels are regulated. With other words, AI Governance needs rules, much more than Internet Governance, where the day-to-day operation could be left in the hands of the technical community without any high risks, as ICANN has demonstrated since the 2016 IANA transition.

This is related to the second difference, which is autonomy. The Internet had a low level of autonomy. Behind the servers, the registries and registrars, there was always an institution that managed the communication, even if some technical procedures were automated. But the level of autonomy in AI has reached a new quality. Take the discussion on autonomous weapon systems (AWS). If a weapon autonomously identifies a person, based on face recognition software, and decides to kill this person, we have a problem. For such decisions, there has to be a human in the loop. In the 21st century, in a rules-based order, where international law (UN Charter) and Human Rights (Universal Declaration of Human Rights) should guide policies, it is against the basic principle of humanity to delegate life and death decisions to machines.

Probably, there are more differences between Internet Governance and AI Governance. But it will be up to the panel to dig deeper into the complexity of the issues and to come up with a reasonable proposal for a comprehensive definition of AI Governance.

Liaisons between the AI-Panel and the IGF MAG

My second proposal is simpler and shorter. What I see is a growing number of discussion places, where all the existing and emerging issues are discussed. And more is around the corner, if I think about Quantum Governance or Neurodata Governance.

On the one hand, this is good. More discussions will bring more ideas and, hopefully, better solutions. But there is a risk of a fragmentation of the discussion landscape. Silos and closed communities could emerge, where the right hand doesn´t know anymore what the left hand is doing. Duplication and a waste of resources could be consequences. And this could become a problem, both for Internet Governance and AI Governance.

We have had the Internet Governance Forum (IGF) for 20 years. The 80th UNGA has now made the IGF a permanent body within the UN System. The Global Digital Compact (GDC), which has created the Global Dialogue on AI Governance, has recognized the IGF as “the primary multi-stakeholder platform for discussion of Internet governance issues”. The IGF has a Policy Network for AI (PNAI). Nearly half of the sessions during the 20th IGF in Lillestrom/Norway in July 2025 were dealing with AI.

Certainly, the IGF is different from the Global Dialogue. But both platforms could benefit from each other if they are able to develop a mechanism for enhanced communication, coordination and collaboration. During the WSIS+20 review, we did hear the call to avoid duplication and to use synergies between existing and new emerging discussion processes. The easiest way would be to connect the two bodies informally via liaisons. The Multistakeholder Advisory Group (MAG) of the IGF could nominate a liaison for the Independent International Scientific Panel on Artificial Intelligence. And the AI-Panel could nominate a liaison for the MAG. This would help to avoid double work and could become a source for additional inspiration.

Thank you very much.