First Global UN AI Governance Summit Wraps Up in Geneva

The UN's first-ever Global Dialogue on AI Governance concluded at the Palais des Nations in Geneva. Over two days, July 6 and 7, representatives from all 193 United Nations member states discussed exclusively how to regulate AI on a global scale, itself an unprecedented event in the organization's history.

Until now, discussions on regulating artificial intelligence have mostly taken place in narrow circles, at forums such as G7 summits, Council of Europe meetings, or closed industry consultations involving the largest tech companies. The Global Dialogue on AI Governance had a different format, open to every UN member state, from the largest economic powers to the smallest developing countries, which until now have rarely had a real voice in shaping global AI rules.

Guterres's Four Priorities

António Guterres closed the proceedings by presenting four areas where, in his view, the world must develop common standards before the next session of the Dialogue. The first is a pact for children's online safety involving AI, requiring developers of these systems to demonstrate their products are safe, to enforce zero tolerance for child sexual abuse material, and to connect children in crisis with human help rather than algorithms alone.

The second priority concerns human rights. Guterres stressed that AI systems must not strip people of their dignity or entrench discrimination, and that decisions in high-risk areas such as the justice system, employment, or healthcare must remain under human oversight. This is a direct nod to concerns raised for months by non-governmental organizations, which warn against automating decisions that affect the lives of millions of people without the ability to appeal to another human being.

Machines can inform, but people must decide and be held accountable for it - António Guterres, UN Secretary-General

Digital Divide and Environmental Footprint

The third priority is narrowing the digital divide between rich and developing countries. More than 20 states backed a proposal to create a global network for AI knowledge-sharing and capacity-building, meant to give developing countries real access to the technology instead of leaving them dependent solely on American and Chinese infrastructure providers.

The fourth point concerns environmental transparency. Guterres called on major AI companies to publicly disclose the carbon footprint, water use, and land use associated with their systems, as part of a new AI environmental transparency initiative. The proposals also included a specific requirement that data centers powering AI systems run exclusively on renewable energy by 2030, which would be one of the first hard, measurable commitments in this area at the global level.

Europe and the Regulatory Sandbox

The European Union, the only bloc of states to have already implemented a comprehensive law regulating AI, presented its own position in Geneva, stressing the need for consistency between national regulations and the global framework being developed by the UN. For Poland and other EU countries, this means future commitments from Geneva will have to be reconciled with the EU's already binding AI Act, whose implementation timeline is itself still disputed in Brussels.

This raises the question of how binding the global rules developed under the UN will actually be in practice, versus how much they will remain a directional declaration with no enforcement mechanism. Critics of the Dialogue point out that the final document contains no sanctions or hard deadlines for implementing the individual priorities, and that its strength will depend entirely on whether individual governments decide to translate the declarations into national legislation.

What Comes Next

Participants in the Dialogue, including representatives of governments, industry, and non-governmental organizations, agreed on one message: the success of global AI governance will depend not on the rules adopted but on concrete actions taken before participants meet again. The second session of the Dialogue will be held in New York in May 2027, and by then individual countries are expected to develop their own positions on the four priorities identified by Guterres.

For tech companies developing AI, including Polish startups building AI-based models and tools, the signal from Geneva means that pressure for environmental data transparency and oversight mechanisms for AI decisions will keep growing in the coming years, regardless of the pace of legislative work in Brussels. Companies already preparing to report the environmental footprint of their models may gain an advantage once similar requirements start applying more broadly.

Sources: UN News (news.un.org), UNESCO (unesco.org), Digital Watch Observatory (dig.watch)