Agenda, Tuesday 16 June (Stockholm)

Reframing Early Warning: Are We Measuring the Right Signals to Enable Preventive Action?

by Early Warning Unit, OHCHR

Most conflict forecasting and crisis early warning systems rely on data of observable manifestations of violence, making them reactive and biased toward data‑rich contexts. While robust, these models often detect crises only after escalation has begun - tracking cycles of violence. To enable earlier anticipatory action, early warning must also capture structural drivers of conflict and root causes, such as socio‑economic trends, discrimination, marginalization, civic space restrictions, political shifts, and climate stressors.

Technological innovation and improved data access now offers opportunities to identify patterns, anomalies, and correlations across a wide range of multiple sources.

The interactive roundtable explores the shift from predicting crises based on symptoms to anticipating them through underlying risk drivers, and the implications for data, methodologies and the integration of human rights analysis. Evidence, including from field experience, shows that earlier preventive engagement is also more effective due to greater political openness and availability entry points.

• Marlene Urscheler, Coordinator Early Warning Unit, Innovation and Analytics Hub, OHCHR
• Adrian Combrinck, Information Management Officer, Emergency Response Team (ERT), OHCHR Bangkok Regional Office
• Corli Pretorius, Head of Research Analytics Learning and Innovation Team, Crisis Bureau, UNDP

Moderator: Alexander Harper, Crisis Insight Team Manager, ACAPS

Seeing Small Signs: Developing a Community-Focused Approach to Violence Early Warning

by The Sentinel Project

Both national and international responses to violence can be slow and limited if they occur at all so vulnerable populations are often left to face threats on their own, acting as ad hoc, self-organized first responders. Members of the public in conflict zones can therefore provide crucial insights that national risk assessments might overlook but they themselves can also lack broader situational awareness that would make their actions more effective. Conflict early warning (EW) systems therefore have great potential to support communal resilience by having a local focus while integrating data gathered from multiple areas and disseminating it to key actors.

SP is developing a framework for localized conflict EW in South Sudan, Benin, and Burkina Faso which involves continuous data collection, analysis, and dissemination of information to support violence prevention and mitigation efforts with community participation being a critical element of the entire process. This session will present the EW framework and related public communication channels while giving participants an opportunity to discuss ways of improving the planned approach and applying it in different contexts.

• Christopher Tuckwood, Executive Director, The Sentinel Project

The LLM as Expert Conflict Coder: A Structured Adjudication Methodology and Its Limits

by Institute for Economics and Peace

Human-coded ordinal indicators remain central to peace and conflict measurement, but human coding is slow, costly, and difficult to scale. Large language models offer an obvious substitute.

This talk presents IEP's structured adjudication methodology for using LLMs as expert coders: defined ordinal scales, detailed coding rubrics, worked country-year anchors, and curated text inputs that together produce scored assessments with justifications. We discuss how the approach is being applied to conflict measurement, and report honestly on what we have learned about its limits.

A workable LLM-coder pipeline has to confront several frictions that are easy to overlook. Models do not natively respect the ordinal structure analysts impose on them. Prompt design choices that look harmless can quietly collapse the output distribution and create false precision. And there are real tradeoffs between reproducibility and useful uncertainty estimation

The aim is to offer a useful map of where LLM adjudication can credibly supplement expert coding, and where it cannot.

• Thomas Morgan, Chief Research Officer, Institute for Economics and Peace

SomReP’s SAADAAL Early Warning System

by World Vision International

SomReP’s SAADAAL Early Warning System responds to a major prevention challenge in Somalia, where climate and hazard risks can quickly escalate into displacement, livelihood loss, resource competition, and local tension when actionable information does not reach decision makers in time. Developed by SomReP consortium, SAADAAL brings together climate and hazard data from multiple sources to generate timely alerts, risk analysis, and decision support for communities, local authorities, and humanitarian actors. By turning complex early warning information into practical and accessible insights, it strengthens anticipatory action, supports locally led preparedness, and informs more timely policy and operational decisions. The session will invite onsite and online participants to compare how communities, local authorities, and humanitarian actors would respond to the same early warning alert, helping surface what data is most useful, for whom, and at what stage of prevention.

• Jerome Galagade, Head of Quality Assurance, M&E and Knowledge Management, Somali Resilience Program (SomReP), World Vision International

Coffee/tea and light refreshments served on the 1st floor

AI and the New Shape of Democratic Power and Peace Processes

by Folke Bernadotte Academy (FBA)

Artificial intelligence is rapidly reshaping the conditions under which democratic governance and peace processes operate. Drawing on FBA’s report "AI and the New Shape of Democratic Power and Peace Processes", this workshop will explore how AI is influencing the dynamics of conflict prevention and peacebuilding.

The session will begin with a presentation of the report’s key findings, focusing on both the risks and opportunities that AI creates for democratic resilience and peace processes. These include the use of AI in elections, public service delivery and citizen engagement, as well as digital dialogues, social media monitoring and ceasefire monitoring. Particular attention will be paid to fragile and conflict-affected settings, where institutional pressures and low public trust can make both the benefits and harms of AI more consequential.

The second part of the workshop will invite participants into a scenario-based discussion set five years into the future. Working in small groups, participants will consider how AI could alter the landscape for democratic institutions and peace processes, including international support. The aim is to identify practical implications for policy and programming, including where AI should be used, limited or avoided, and what capacities institutions and partners need to build now.

Speakers:

• Catalina Perdomo, Senior Strategic Officer, Support to Peace Processes, Folke Bernadotte Academy (FBA)
• Emma Skeppström, Deputy Director Conflict Prevention and Peacebuilding Department, Folke Bernadotte Academy (FBA)
• Dr. Martin Wählisch, Associate Professor of Transformative Technologies, Innovation and Global Affairs, University of Birmingham

Closing the Action Gap: From Signals to Systems for Anticipating the Impacts of Fragility, Conflict and Violence

by Danish Refugee Council

Early warnings for conflict, fragility and violence are becoming more credible and timely. Communities respond to them. Analysts model them. Yet humanitarian systems still struggle to act before crises escalate. The challenge is no longer information alone, but the political, financial and operational commitments required to act under uncertainty.

While anticipatory action (AA) has advanced for climate shocks through pre-arranged financing, agreed triggers and tested delivery systems, equivalent systems for fragility, conflict and violence remain limited. This leaves millions exposed to preventable compounding impacts, including forced displacement, food insecurity, protection risks and livelihood collapse.

This session will explore what it takes to build AA systems that can operate across FCV impacts. Panellists from local organisations, regional institutions, operational agencies and donors will examine conflict-sensitive triggers, financing mechanisms that release resources before crises peak, and the institutional risk tolerance needed to act on probabilistic signals.

• Morten Schacht Högnesen, Head of Section, Innovation, Data & Analysis, Danish Refugee Council (DRC)
• John Jal Dak, Executive Director, Youth Social Advocacy Team (YSAT)
• Antje Lehmann, Fund Manager, Complex Risk Analytics Fund (CRAF’d)
• Marek Baron, Research Analyst, International NGO Safety Organisation (INSO)
• Jerome Galagade, Head of Quality Assurance, M&E & Knowledge Management at Somali Resilience Program (SomReP), World Vision International

Moderator: Anna Lena Huhn, Global Lead on Anticipatory Action, Danish Refugee Council (DRC)

Displacement Signals in Active Conflict. eRADAR: Emergency Regional Analytics for Displacement Anticipation and Response

by UNHCR

Active conflict makes displacement hard to measure. Borders close, field access narrows, and official counts arrive weeks late. Response decisions get made before the numbers do.
eRADAR is a displacement nowcasting pipeline built for that window. The method anchors detection to confirmed conflict events, cross-validates against multiple remote sensing sources, and produces nightly movement estimates across 100+ official and unofficial border crossings and 17+ countries in the Middle East and North Africa.
The contribution is nowcasting, not forecasting. eRADAR estimates the present, the days when decisions are made and field data does not yet exist, complementing the monthly forecasting cycles many have built.
This talk walks through the method: what it gets right, what it is not, and how to use it in active conflict. We trace one conflict event through to a corridor-level movement signal using a recent detection case, and try to answer the working question: given confirmed events up to yesterday, where would you expect people to move today?

• Md. Rifat Zahir, Data Scientist, UNHCR
  

Improving Access to Evidence: Introducing UNIDIR's Interactive Data Dashboard on Exits From Armed Groups

by United Nations Institute for Disarmament Research (UNIDIR)

In this demo, the Managing Exits from Armed Conflict Project at the United Nations Institute for Disarmament Research will demonstrate its interactive data dashboards. The MEAC project generates unique evidence through quantitative and qualitative data collection on individual journeys into and out of armed groups and the experiences of conflict affected populations, helping practitioners design interventions to prevent recruitment and support exits after conflict involvement.

To provide practitioners and researchers with more direct access and engagement with the quantitative data collected by the project in Iraq, Colombia, North West Nigeria and the Lake Chad Basin region, a selection of the project’s surveys with conflict-affected populations have been released to the public on data dashboards. These include survey data on those formerly associated with armed groups (or those perceived as such), as well as unaffiliated community members in conflict-affected areas.

• Dr. Jente Marit Althuis, Researcher, Data and Methodology Lead - United Nations Institute for Disarmament Research, Switzerland

Lost in 90 Tabs: Closing the Last Mile Between Humanitarian Data and Field Decisions

by Norwegian Refugee Council

This session presents CLEAR, an initiative by the Norwegian Refugee Council, building an open-source digital backbone that ingests crisis data from conflict monitors, media, weather services, displacement trackers, community reports, and field officer inputs. The system uses advanced analytics and AI/ML to cluster fragmented signals into coherent events, enriched with situational awareness of needs on the ground, and delivers actionable alerts to responders and partner platforms through a single open API.

• Zineb Bhaby, AI Lead, Norwegian Refugee Council (NRC)

Lunch served on the 1st floor.

The Data We Need to Invest in Peace

by DPPA-DPO Information Management Unit and the Digital Transformation Strategy of Peacekeeping

The New Agenda for Peace and UN80 articulated a cost-benefit analysis of conflict prevention. However, Member State investments in conditions for peace don't always follow. At the same time, data maturity of conflict studies has increased, with measurable dimensions such as conflict intensity, duration & spread readily available. Experts across health, economics and education can rapidly trace the conflict burden within their domains, but policymakers may not yet be proficient in data-informed advocacy on conflict issues.

This panel seeks to strengthen the value-proposition of investing in peace by exploring a synthesis of conflict forecasting and economic modelling and showing how evidence makes the case for action – i.e., including in clarifying the costs of inaction and the consequences of different policy choices. The discussion helps identify new advocacy strategies – linking conflict forecasting to consequences at regional, global, and conflict-party levels for individual conflicts.

The panel brings together expertise – from economics, conflict forecasting and policy domains – to answer, “what role evidence and advocacy could play to strengthen investments in peace?"

• Dr. Simon Polichinel von der Maase, Lead Data Scientist and Director of Technical Development, VIEWS, Peace Research Institute Oslo
• Dr. Astrid Irrgang, Executive Director, ZIF - Center for International Peace Operations
• Anayansi Lopez, Head of Pillar – Human Rights, Civil Affairs, Peacebuilding, Guidance and Learning, Policy and Best Practice Service (PBPS), Division of Policy Evaluation and Training (DPET), Department of Peace Operations (DPO), United Nations
• Thomas Morgan, Chief Research Officer, Institute for Economics and Peace
  

Moderator: Ayham Al Maleh, Information Management Officer, DPPA-DPO Information Management Unit, United Nations

From Conflict Forecasts to Socio-Economic Impacts: Operational Use Cases and Challenges

by Department of Peace and Conflict Research, Uppsala University, and Peace Research Institute Oslo

Timely information on the socio-economic impacts of conflict is limited. Key indicators such as GDP per capita and child mortality are often reported only annually and with delays, reducing their usefulness for early warning and proactive decision-making.

We present a machine learning framework that forecasts these indicators at monthly resolution using high-frequency conflict data. We produce forecasts up to 48 months ahead at both country and subnational (PRIO-grid) levels, with an end-to-end pipeline including data integration, model evaluation, and nowcasting to address reporting delays.

The talk will show how this approach translates predicted changes in conflict intensity into anticipated socio-economic deterioration. Using selected retrospective and forward-looking cases, we illustrate how the model can signal impacts earlier and support early warning, monitoring, and operational planning. We also briefly demonstrate API-based access and outline ongoing work to expand indicators and use cases in collaboration with partners.

• Borbála Farkas, Research Engineer, Department of Peace and Conflict Research, Uppsala University

The High-Frequency Pulse: Fusing Ground Truth and GenAI for Real-Time Mission Intelligence

by Premise Data

Problem: UN missions operate in fluid environments but are hampered by a verification crisis. Current ground-truth methods—reliant on manual, physical deployments—are slow, complex, and expensive. This forces a reliance on anecdotal evidence or post-crisis analysis. Without high-frequency ground-truth loops, missions cannot adapt at the speed of the crisis.

Innovation: This session explores remote monitoring by fusing validated ground truth with GenAI. Using Premise’s AEGIS platform, we show how organizations establish continuous insight loops. By coupling proven ground-truth validation with a GenAI engine, AEGIS unifies community insights with OSINT for a real-time pulse on sentiment. This remote-first approach drastically reduces costs while providing ongoing situational awareness. Agile, AI-driven loops ensure peacebuilding is proactive rather than reactive.

• Behar Xharra, Director, International Development and Humanitarian Assistance, Premise Data

Democratizing High-Resolution Mapping: Community Drone Data for Inclusive and Conflict-Sensitive Decision-Making

by Humanitarian OpenStreetMap Team

This session presents how HOT is democratizing high-resolution geospatial data through open, community-driven drone mapping.

We will showcase HOT’s open drone infrastructure, enabling local pilots to safely capture imagery (up to 2 cm GSD), coordinate mapping across complex environments, and openly process and share data. By lowering barriers through affordable tools, simple workflows, and local capacity building, communities can document their environments, generate evidence for advocacy and services, and support more inclusive decision-making.

This session will include a demo and case studies from Freetown and Lusaka illustrating community-driven drone mapping of 80–100 km² in informal settlements. Ultimately, this work contributes to a broader vision of people-centered, open data ecosystems that support peace, resilience, and local agency.

• Ping Yee, Lead Software Engineer, Drone Services, Humanitarian OpenStreetMap Team

Coffee/tea and light refreshments served on the 1st floor

Signals to Solutions: A Live Data-to-Action Simulation on Syria

by The What If gGmbH

The What If’s (TWI) interactive lightning talk offers a hands-on demonstration of Comprehentia HDP, our conflict-sensitive narrative intelligence system that bridges large-scale data analysis with real-time peacebuilding action. We address a key challenge in the Data for Peace field: while vast amounts of data are collected, they often remain disconnected from locally actionable responses. Session-participants will engage in a rapid “data-to-action” pitch using a live simulation to explore emerging narratives, detect risks such as polarisation, misinformation, and emotional amplification (e.g. spikes in fear or anger), and propose response strategies. Through live polling and guided interaction, the session walks participants through a clear pathway: from raw data to narrative insight to concrete intervention. Building on our implementation in contexts such as Syria, the session demonstrates how narrative and sentiment analysis can be operationalised for early warning and prevention systems. Participants will actively engage in interpreting data and co-designing responses, simulating how practitioners, policymakers, and civil society can apply these insights in practice.

• Madeline Wiedemann, Head of Peacebuilding, The What If

Early Warning in Action: Leveraging Data and Technology for Violence Prevention

by CMI – Martti Ahtisaari Peace Foundation

Policymakers, academics, and peacebuilders are increasingly integrating data and emerging technologies in their work to support more timely and effective responses to violence in all its forms. At the same time, organizations are integrating data-informed early-warning systems differently within their decision-making processes. This session brings together leaders in the field to examine how early-warning systems are embedded within different institutions, what it takes to make data-informed prevention operationally useful, what the limitations and trade-offs of current approaches are, and how institutions and practitioners are adapting to evolving policy and operational realities.

The panelists will explore the tools, methodologies, and institutional approaches they employ to integrate data and technology into conflict prevention, monitoring, and foreign policy. The discussion will touch on themes related to institutional uptake, operational integration, local ownership, sustainability, and cross-sector collaboration.

By connecting scientific innovation with operational experience, the panel aims to identify practical pathways for strengthening collaboration across sectors, strengthening the role of data and technology for decision-making processes, and advancing more responsive, inclusive, and context-sensitive approaches to violence prevention and early action.

• Dr. Marius Hofmann, Desk Officer for Crisis Early Warning, German Federal Foreign Office
• Aurora Bergmaier, Policy Officer for Early Warning and Conflict Prevention, EEAS
• Dr. Olli Ruohomäki, Senior Advisor, Political Department, Ministry of Foreign Affairs for Finland
• Fraser Alcorn, Conflict Analysis Lead, United Kingdom Foreign, Commonwealth & Development Office (UK FCDO)

​Moderator: Johanna Poutanen, Head of Inclusion and Digital Innovation, CMI – Martti Ahtisaari Peace Foundation (moderator)

Closing Remarks

by The Data for Peace Organizing Committee

Acknowledgements and closing remarks concluding the first two days of the 2026 Data for Peace Conference.

• Angelica Lindqvist-McGowan, Director of Operations & Outreach – Violence & Impacts Early-Warning System (VIEWS), Department of Peace and Conflict Research, Uppsala University