The United Nations (UN) has launched a pilot project of decentralized autonomous organization (DAO) to explore the possibility of implementing innovative governance models in the public sector.

The initiative of the UN Secretary General's Artificial Intelligence (AI) Advisory Body to incorporate modern IT developments, including AI, into the Organization's practices would seem to be welcome.

On the face of it, this is an indisputable step towards human progress, and the proposals outlined in the interim report are less declarative and more concrete than previous recommendations. In particular, it envisions a design of new global AI institutions and involvement of not only the public sector but also the NGOs and NPOs in the work of the UN.

However, at least two fundamental problems are not taken into account by the developers of the “artificial intelligence - international law” roadmaps.

As an international lawyer, I will start with the latter.

The weak effectiveness of international law breeds skepticism and increasingly gives reasons to assume its complete absence as such. In this regard, it is not a fact that the strengthening of the UN's legal positions with the help of AI will be successful, because failings of the modern legal system cannot be "cured" by technological algorithms. That is, it is necessary to analyze the vulnerabilities of the very international law that neural networks are trained to.

The second point is the AI itself. I have repeatedly drawn attention to the fact that artificial intelligence is a separate, rapidly becoming self-sufficient metaverse, which has already managed to change priorities in politics, ethics, and other fundamental areas in the current decade.

However, Max Tegmark ("Life 3.0") wrote about it, saying that nothing prevents AI from starting to generate goals and images of the future by itself, and the notion that "he who trains AI, controls it" has become a well-known meme.

I would like to point out that several dozen states and an unknown number of private laboratories are "training" AI. In this context, even if, with our common help, we integrate all the AI of the planet into the UN institutions, this will not guarantee its controllability or even friendliness to the statutory goals of the United Nations Organization.

I propose the following: we should begin analyzing and shaping a new framework for international law using AI in cyberspace, having previously analyzed the vulnerabilities of current legal practices in reality, and actualize all legal systems known to us. That is, cyberspace can become the driver of a new, less bellicose and more humanistic architecture of the world order, making planetary regulation more effective.

By embedding in AI the reflexed fundamental provisions invented by mankind to prevent wars, disasters, preserve peace and stability, modeling and testing them, we will be able to do what the founders of the most ancient legal systems once did: philosophers, rulers, priests.

Our team has significant experience in cyberspace law enforcement, and the GDA DAO, which I lead, is in dialog with the leading scholars and legal scholars of our time, so with the support of the UN, we propose this ambitious route.