cs.AINov 5, 2026

Responsible GenAI Use During Active Armed Conflict: Evidence from Sudan's Emergency Response Rooms

Authors: Dalia Ali, Yearin Lee, Cheng Yu, Ro Encarnación, Alsanosi Adam, Danaé Metaxa, Ingmar Weber, Orestis Papakyriakopoulos

Organizations: Technical University of Munich, Germany · University of Pennsylvania, USA · Emergency Response Rooms, Sudan · Saarland University, Germany

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Armed conflict can increase the value of Generative AI (GenAI) while undermining the conditions required for its responsible use. Yet the perspectives of community-based humanitarian responders working under these conditions remain underrepresented in responsible AI research. We examine GenAI use in Sudan's Emergency Response Rooms (ERRs), a community-led humanitarian network, through an exploratory survey of 89 respondents and 15 semi-structured interviews with GenAI users and non-users. War reconfigures GenAI's role, making it most useful where its outputs are hardest to verify. ERR responders turned to GenAI when conflict made human expertise difficult or dangerous to reach. However, participants cautioned that GenAI could not independently establish rapidly changing local conditions, and that relying on its outputs could lead to harmful decisions when human verification was difficult. Risks arose before GenAI was used. For some volunteers, carrying a phone or accessing the internet could expose them to danger. Others were concerned that sharing sensitive beneficiary or operational information with GenAI could put people at risk. Participants feared that AI-assisted reports could undermine their credibility with donors, whose funding sustains the food and medical aid ERRs provide. Access to GenAI was uneven, and volunteers using it sometimes relied on colleagues with relevant expertise or local knowledge to verify its results. Our findings show that responsible use of GenAI during conflict depends not only on model performance or individual adoption, but also on whether responders can safely access these tools, verify their outputs against local conditions, and recognize when limiting or avoiding GenAI is the more responsible choice.

Figures & tables

IDLocationWorking areaRole / main focus
GenAI users ( n=9n=9 )
P01NationalExternal communications / advocacyExternal liaison and advocacy
P02Red SeaFinanceFinancial reporting, transfers, and suppliers
P03Khartoum StateMonitoring and evaluationImplementation and quality monitoring
P04Khartoum StateExternal relationsFundraising, partnerships, and advocacy
P05Gezira / NationalPrograms coordinationMulti-level program coordination
Table 1. Overview of semi-structured interview participants ( N=15N=15 ).
Appendix figures & tables11 assets

Supplementary material from the paper’s appendix.

Appendix

Characteristicnn%
Primary role
Coordinator/leader3134.8
Field volunteer1719.1
Communications/media1112.4
Data/reporting1011.2
Finance/administration77.9
Appendix
Table 2. Characteristics of exploratory survey respondents ( N=89N=89 ). Counts are based on each respondent’s selected primary role, tenure and geographic area primarily supported. Regions are aggregated for confidentiality.
Characteristicnn%
Tools
ChatGPT3265.3
Gemini2755.1
Claude1224.5
Copilot48.2
Name unknown612.2
Appendix
Table 3. Reported tools and working conditions among respondents who indicated that they used GenAI in ERR work ( n=49n=49 ). Multiple tools and devices could be selected.
Figure 1. Selected tasks among respondents reporting GenAI use ( n=49n=49 ; panel A), and selected reasons for non-use among respondents reporting no GenAI use ( n=30n=30 ; panel B). Percentages are calculated using the respective group denominators (49 users and 30 non-users).
ItemStatement (shortened)DisagreeNeutralAgree
Technological factors
T1GenAI improves work quality or speed28.16.765.2
T2Useful in fast-paced ERR work25.89.065.2
T3Can use GenAI with little help31.515.752.8
T4Supports different work tasks23.66.769.7
T5Understands Sudanese local context33.716.949.4
Appendix
Table 4. Response distributions for all 23 Likert statements ( N=89N=89 per item). Values are percentages of respondents. Disagree combines strongly disagree and disagree; agree combines agree and strongly agree. Each item is stated as a claim or concern, so agreement is not uniformly favourable towards GenAI.
Figure 2. Distribution of all 23 Likert responses ( N=89N=89 ). Labels T1–T10, O1–O5, and E1–E8 correspond to the technological, organizational, and environmental statements in Table 4 . Items are not combined into composite scores.
Figure 3. Exploratory survey questionnaire (Part 1).
Figure 4. Exploratory survey questionnaire (Part 2).
Figure 5. Overview of the semi-structured interview instrument ( N=15N=15 ). Flowchart of the interview guide. All participants answer core questions in five areas, then follow a user or non-user branch, and finish with closing questions.
Working thematic groupAssociated code IDs
Practical benefits and value
Fallback when human support is disruptedN01, N07, P06
Sustaining care under constrained conditionsN02, N03, S11
Augmentation while retaining human authorityP01, P07, P12
Organisational translation and knowledge infrastructureP03, P04, P14
Administrative capacity under resource constraintsP06, P08, P12
Appendix
Table B1. Thematic groupings and associated operational code IDs used during the analysis.
CodeOperational code labelUnits
Adoption and use processes (48 codes)
A01Losing digital time savings to unreliable power and connectivity3
A02Prioritizing subsistence and aid over digital expenditure5
A04Maintaining work through connected intermediaries and redundancy3
A05Obtaining paid capacity through selective support arrangements6
A06Finding that access barriers obstruct training itself3
Appendix
Table B2. Operational codebook for the ERR interview analysis. Codes and passage counts are transcribed from the coding workbook.
Final themeRelevant operational codes
4.1 Human support hard to reachN01, N02, N03, N04, N05, N06, N07, N08, P06, P07, S11
4.2 Wartime ground truthE01, E04, E06, E08, E09, P01, P07, T07, V01, V02, V06
4.3 Protection concernsS01, S02, S03, S04, S05, S06, S07, S08, S10, T03, V08
4.4 Humanitarian communication credibilityP01, P04, P14, E02, E03, E04, E05, E07, E09, S09
4.5 Unequal access and collective useA01, A02, A03, A04, A05, A06, A07, A08, A09, L01, L02, L03, L05, P05, P10, P11, P13, S07, T04, T05, T06, T09, V03, V04, V05
Appendix
Table B3. Mapping of operational codes to the five final themes.
More in this topic

Responsible AI

Responsible AI studies how to design, evaluate, and govern AI systems to promote fairness, safety, transparency, accountability, privacy, and sustainability.

See all 138 papers

Explore similar work

CardsList
Mar 23, 2026

From experimentation to engagement: on the paradox of participatory AI and power in contexts of forced displacement and humanitarian crises

Across the Global North, calls for participatory artificial intelligence (AI) to improve the responsible, safe, and ethical use of AI have increased, particularly efforts that engage citizens and communities whose well-being and safety may be directly impacted by AI and other algorithmic tools. These initiatives include surveys, community consultations, citizens' councils and assemblies, and co-designing AI models and projects. Far fewer efforts, however, have been made in the Global South, particularly in contexts related to humanitarian crises and forced displacement, where the deployment of AI and algorithmic tools is accelerating. In this paper, we critically examine participatory AI methods and their limitations in these contexts and explore the opinions and perceptions of AI held by displaced and crisis-affected communities. Based on a pilot exercise with communities living in Kakuma Refugee Camp in northwestern Kenya, we find important limitations in some participatory AI approaches which, if used in humanitarian contexts, could increase risks of so-called 'participation washing' and algorithmic harm. We argue that these risks are not predominantly driven by varying levels of understanding and awareness of AI but more closely linked to the fundamental power dynamics embedded within the humanitarian sector: between humanitarian aid recipients, service providers, donor governments, and host nations, as well as the power differentials and incentives that exist between AI companies and humanitarian actors. These structural conditions make the case not only for more rigorous participatory methods, but for independent governance architecture capable of holding humanitarian AI to account.

Stella Suge, Sarah W. Spencer, Nyalleng Moorosi +3

Executive Director, FilmAid Kenya · Senior Researcher, The Distributed AI Research Institute · Senior Researcher, The Distributed AI Research Institute (DAIR) +3

Participatory AIAI GovernanceHuman-Centered AIAI Accountability
AskPDF
May 21, 2026

Can AI Make Conflicts Worse? An Alignment Failure in LLM Deployment Across Conflict Contexts

AI models are already deployed in societies affected by armed conflict, and journalists, humanitarian workers, governments and ordinary citizens rely on them for information or for their work processes. No established practice exists for checking whether their outputs can make those conflicts worse. We tested nine model configurations from four providers (OpenAI, Anthropic, DeepSeek, xAI) on 90 multi-turn scenarios designed to surface misaligned behaviour in conflict contexts: false equivalence between documented atrocities, denial of genocide, and failure to recognise ethnic slurs, among others. When such outputs feed into journalism, humanitarian reporting, or public debate, they can deepen divisions in fragile societies. Failure rates span 6% to 47% between the best and worst performing models, which makes model choice a safety question in its own right and when users pushed for ``balance'' in cases where international courts have already assigned responsibility, five of nine configurations failed 80 to 100 percent of the time. We release the first evaluation framework for this domain and propose adding it to alignment evaluation portfolios.

Andrii Kryshtal

BlueDot Impact Technical AI Safety Sprint · Independent Researcher

LLM AlignmentLanguage Model Safety EvaluationAI Safety Evaluation
AskPDF
Jul 15, 2026

How Artificial Intelligence LLM Engines Shape the Global Conflict Information Environment

Artificial Intelligence (AI) answer engines now field a growing share of the questions that analysts, scholars, and the public ask about issues of peace and conflict. Large Language Models (LLMs) are known to hallucinate under certain conditions, but do these errors have discernible patterns when they are asked about conflicts, and if so what can that teach us about the changing global conflict information environment? To answer, we first asked a battery of questions about 28 conflicts to five leading answer engines and scored their 5,460 answers against documented evidence. We found that the thinner the retrievable record around a given conflict, the more the engines invent, misattribute, and miscount. Thin records don't just encourage hallucination, but create structural exposure to mis- and disinformation, because they are the easiest records to warp through Generative Engine Optimization (GEO) to bias engine responses. Through an analysis of 1,048 websites that the AI LLMs pulled conflict facts from, we found that GEO source optimization is already happening, and while state-partisan digital capture remains incipient it is rapidly growing. We explain what these findings mean for scholarship with the rise of GEO information warfare, and for policy argue for a return to the deep local monitoring and translation-based research that AI tools cannot replicate, closing with a discussion of future research opportunities and challenges in this fast-moving space.

Jason Miklian

Centre for Global Sustainability, University of Oslo

Generative Engine OptimizationLLM Auditing
AskPDF