President of General Assembly attends Global Dialogue on AI Governance in Geneva, July 6, 2026. UN Photo/Irina Popa.

The first Global Dialogue on AI Governance, convened in Geneva on July 6 and 7, 2026, is the UN’s flagship multistakeholder forum on artificial intelligence (AI). Together with its companion body—the Independent International Scientific Panel on AI—it forms the institutional core of a governance architecture rooted in the Pact for the Future and the Global Digital Compact (2024) and formally authorized by General Assembly Resolution 79/325 (2025). The Panel produces annual evidence-based assessments of AI’s opportunities, risks, and societal impacts. The Dialogue translates that evidence into political conversation among governments, civil society, and the private sector. Neither body carries binding authority, but both hold the power to decide what qualifies as an AI governance issue. Yet the intersection of AI with peace has not yet found its place on the agenda, despite peace being the very reason the UN exists.

In her opening remarks at the Dialogue, Annalena Baerbock, President of the General Assembly, turned to the UN Charter—specifically its language on international cooperation on problems of “an economic, social, cultural, or humanitarian character”—to answer those questioning whether AI belongs on the UN’s agenda. But invoking the Charter as the source of the UN’s authority to govern AI means invoking the whole of it. The UN’s overriding purpose, stated in the Charter, is “to save succeeding generations from the scourge of war.” That first commitment cannot be quietly left out. The UN was forged in catastrophe to prevent catastrophe. If its newest governance instruments cannot hold space for that original purpose, they are incomplete.

Military Uses of AI Have Been Conflated with AI and Peace

The General Assembly resolution explicitly limits both the Panel and the Dialogue to “the non-military domain,” leaving military AI to separate tracks. Because military AI was excluded, peace appears to have been excluded alongside it, as though the two were inseparable. They are not. Civilian AI intersects with peace in two directions. On the one hand, it can support mediation, atrocity prevention, and human rights documentation. On the other hand, it also amplifies hate speech, facilitates repression and surveillance, and destabilizes political transitions. These intersections between AI and peace are nonmilitary; excluding issues like lethal autonomous weapons systems does not require excluding peace.

The resolution does not mention peace by name. But neither does it mention health or education, both domains that have been discussed extensively because they sit at the intersection of AI’s capacity to help and to harm. The resolution additionally frames the Dialogue as a platform to advance the Sustainable Development Goals (SDGs). SDG 16 on peace, justice, and strong institutions holds the same claim to inclusion as SDG 3 on health or SDG 4 on education. What is absent is not a legal basis but the political will to act.

The reason is likely political. Discussions around peace can be more politically sensitive, particularly among states with divergent security interests. The nonmilitary limitation was probably a compromise to keep major powers at the table. For the Dialogue, where political consensus must be negotiated, such caution is understandable, even if not defensible. But the Panel was built on a different premise—and it is there that the silence is hardest to explain.

Peace Is Absent from the Panel’s Preliminary Report

The Panel’s Preliminary Report, released ahead of the Geneva convening, describes itself as a balanced assessment of AI’s opportunities and risks, yet nowhere does it reference peace. That absence may not mean it went unconsidered: at the report launch, co-chair Maria Ressa noted that members “fought” and called the report a floor, not a ceiling, suggesting peace may have been discussed but did not survive the negotiation into the final consensus text.

The report devotes an entire chapter to AI’s effects on human rights, information integrity, and democracy. It documents how AI-generated deepfakes and persuasive content weaken the shared informational foundations on which public trust, social cohesion, and democratic deliberation depend. It identifies how AI-assisted cyberattacks, disinformation, and the concentration of AI capabilities could facilitate authoritarian capture and erode democratic accountability.

The report thus engages the substance of peace without ever naming it. The erosion of shared reality creates an environment where conflict narratives can take root, intergroup violence can escalate, and mass atrocities can become more likely. Authoritarian capture is a precondition for systematic repression. Labeled only as “information integrity” or “democracy” rather than as peace issues, these harms are not linked to conflict analysis, early-warning systems, and atrocity-prevention frameworks. The report also does not reference conflict-affected settings or engage the substantial evidence base on AI for peace. What is not named is not measured, resourced, or acted upon. The Panel’s next iteration will inform the second Global Dialogue (2027) and eventually the Global Digital Compact review (2027–2028), providing a window for correcting this gap.

Unlike the politically constrained Dialogue, the Panel is an independent scientific body meant to document evidence and flag knowledge gaps, making its omission of peace harder to justify. What it omits, it omits not for lack of consensus among states but against the weight of the evidence. A substantial and growing body of research already documents AI’s impacts on peace—from conflict forecasting and atrocity prevention, to algorithmic amplification of hate speech in conflict-affected societies, to AI-assisted negotiation and mediation tools. This literature exists, and it is directly relevant to the Panel’s mandate. A scientific body that omits evidence because the political framing is contested is not exercising caution; it is demonstrating a lack of independence.

The Dialogue’s Agenda Did Not Include Peace

Like the Panel’s preliminary report, the Dialogue agenda also omitted peace. The Dialogue was organized around four thematic clusters: AI opportunities and implications; bridging AI divides; safe, secure, and trustworthy AI; and respecting, protecting, and promoting human rights. Peace appeared in none of them. Yet from the opening speeches onward, it kept finding its way into the room.

The secretary-general set the tone. He acknowledged that AI respects no civilian–military boundary: the same models and chips used by civilians are also being used on battlefields. He named lethal autonomous weapons systems as his primary concern, urging us to “call them what they are: killer robots.” Baerbock cited a Rwandan teenager who warned that “poison” now travels faster, “hiding in screens… whispering old lies to a new generation.”

While only a few representatives of conflict-affected states delivered high-level statements, several of those that did could not stay silent on peace. Libya’s representative called for a binding legal framework regulating the use of AI in armed conflict, while Iran raised ethical and legal concerns about AI, including its “alarming misuse” in conflict. The Democratic Republic of the Congo insisted on the need to combat conflict minerals in the AI supply chain.

Despite these opening statements, peace remained absent through most of the Dialogue. Cluster 4 on human rights was where peace most naturally belonged. Human rights and peace are structurally inseparable—human rights violations are among the earliest indicators of conflict escalation, and their protection is foundational to atrocity prevention. Yet even here, peace was not named, conflict-affected settings were not referenced, and the evidence base on AI’s role in conflict prevention, mediation, and civilian protection was not engaged.

Peace Finds a Voice on the Margins

Some expected that if peace did not find a place at the AI Dialogue, it would still find a place at Geneva Digital Week through the AI for Good Global Summit. Though the Summit introduced a dedicated UN AI Hub Stage showcasing AI applications in human rights and humanitarian emergencies, peace was once again absent. This omission is especially perplexing considering that the UN’s own AI-enabled tools for early warning and anticipatory action are among the technology’s most consequential civilian applications.

Peace is not merely absent from these conversations but structurally excluded. At the AI for Good Global Summit In 2025, anti-war activists who disrupted a main-stage speaker over corporate responsibility in military AI were physically removed, and a keynote speaker’s references to Palestine and genocide were censored. This has not stopped activists from pushing these issues into the spotlight. In 2026, activists organized the People’s AI Summit as a counter-event, focusing on the role of tech companies in enabling mass surveillance and war crimes. Meanwhile, activists disrupted the AI for Good Summit with “No Tech for Apartheid” banners at the Amazon stage, drawing attention to AI-powered targeting systems deployed in Gaza. They exposed a contradiction within the UN architecture: part of the system produces rigorous evidence on business and human rights obligations, while UN summits platform the very corporations implicated in violating those standards.

Filling the Gap before the Next AI Dialogue

The first AI Dialogue was a significant institutional accomplishment, and its organizers deserve recognition. But when it comes to peace, the gap is real, consequential, and correctable. The UN was established to prevent war and build peace. Eighty-one years later, its flagship AI governance mechanism does not address the civilian peace applications falling squarely within its mandate—not because evidence is lacking, practitioners are absent, or the institutional infrastructure is missing but because a political compromise designed to exclude military AI inadvertently excluded peace alongside it.

The path from Geneva to the next AI Dialogue in New York offers a finite but genuine window to correct this. The Panel can integrate the evidence, the Dialogue can open the agenda, and the practitioner community can organize and submit—none of which requires renegotiating the resolution or breaching the nonmilitary boundary. The evidence and institutions are ready. The agenda must follow.

For the AI Panel, three practical steps would allow it to bring “AI for peace” expertise into its work without exceeding its current mandate. The Panel could begin by issuing a formal call for submissions from AI for peace researchers and practitioners, inviting them to identify the evidence, risks, and applications it should address; the founding resolution already authorizes informal consultation with external experts, but a structured intake anchored in Working Group 5 would surface the field’s priorities systematically rather than leaving them to chance. From those submissions, the Panel could synthesize an interim thematic brief on AI and peace and incorporate peace into its next report—a step that requires no expansion of authority, since the Panel’s methodology already provides for such briefs, grounded in its mandate to assess AI’s “opportunities, risks and impacts.” A briefing workshop during the drafting period could then connect Panel members directly with contributing experts, ensuring the evidence informs the report while it is still being written and embedding the field’s expertise into the Panel’s deliberations rather than leaving it to be considered after the fact.

For the AI Dialogue, a parallel set of moves would translate the Panel’s work into political conversations. Peace and conflict prevention could be established as a formal agenda item, carrying the Panel’s thematic brief into the co-chairs’ “priority areas”; the founding resolution already mandates the advancement of the SDGs, so no new mandate is required—only the decision to use it. Alongside a formal agenda placement, standing participation for AI for peace practitioners as formal speakers and participants, connected through a coordination mechanism before and during the Dialogue, would make engagement continuous rather than episodic. And where the formal agenda moves slowly, side events, parallel convenings, and community-led sessions can advance these connections independently—engaging Dialogue speakers and audiences on AI and peace through the surrounding ecosystem, not only through the Panel and the Dialogue.

Branka Panic is Founding Director of AI for Peace.