Balancing Benefits, Risks, and Governance for Responsible Data Sharing in Humanitarian Action
Previously in our Data Sharing series, we discussed Responsible Data Sharing: Balancing Promise and Risk.
Introduction
In humanitarian work, data has become as critical as food, water, or shelter. Whether it is tracking the spread of cholera, mapping food insecurity, or coordinating cash transfers, decisions rise and fall on the quality and availability of information. Yet data in this sector is not neutral: it carries the identities, vulnerabilities, and sometimes even the survival prospects of those we aim to serve.
That tension, between the power of data to save lives and the risks of exposing the people behind the numbers, is at the heart of today's humanitarian operations. Organizations such as the World Food Programme (WFP), FEWS NET, FAO, GiveDirectly, the Humanitarian Data Exchange (HDX), and newer innovators like the Global Development Incubator (GDI) are all grappling with the same challenge: how to share data in ways that enhance coordination and accountability without undermining trust or compromising safety.
As someone working at the intersection of data analysis and governance, I see this not as a technical detail but as a defining issue for the sector. The question is no longer whether to share data, but how to share it responsibly.
This article examines how humanitarian organizations are sharing data responsibly, maximizing its value for good while safeguarding against misuse, and outlines a practical governance framework to ensure data is managed ethically and effectively. It draws on practices from WFP, UNHCR, OCHA's Humanitarian Data Exchange, GiveDirectly and others, and positions responsible data governance as a critical skill set for humanitarian professionals and policy-makers alike.
The Promise of Data Sharing in Humanitarian Work
Effective data sharing can dramatically improve humanitarian outcomes by breaking down silos and enabling collective intelligence. The key benefits include:
Improved Coordination and Efficiency
When agencies pool their information, they can identify needs and gaps more quickly, avoid duplication, and deliver aid more efficiently. In Tanzania's refugee camps, UNHCR and WFP connected their systems through the PRIMES Interoperability Gateway (PING), which automates a previously tedious, manual exchange of refugee registration data and allows data to flow in both directions. Because PING draws directly from UNHCR's registration database, WFP can plan assistance on the most up-to-date household data, and staff on both sides report better coordination and more efficient data management (UNHCR).
Transparency and Accountability
Open data fosters trust among stakeholders: donors and affected communities alike can see where resources are going. OCHA's Humanitarian Data Exchange (HDX) now carries more than 20,000 datasets shared by over 200 organizations, with users in 230 countries and territories (OCHA Centre for Humanitarian Data). Crucially, HDX balances openness with safeguards: sensitive datasets are shared through HDX Connect, where only the metadata is public and the underlying data is released on request (OCHA Centre for Humanitarian Data). Transparency and protection can coexist.
Innovation and Insight
Data sharing also fuels innovation. During the COVID-19 pandemic, GiveDirectly and researchers at UC Berkeley helped the Government of Togo combine satellite imagery and mobile phone metadata to identify the poorest people for emergency cash transfers. A peer-reviewed evaluation in Nature found the machine-learning approach reduced exclusion errors by 4 to 21% compared with the geographic targeting options the government was considering, while noting it would be less accurate than a comprehensive social registry, had one existed (Aiken et al., Nature, 2022). Combining data from telecom operators, governments and NGOs can reach people faster when traditional data is missing or out of date.
Risks and Challenges: Privacy, Consent, and Context
While the upsides are significant, humanitarian data sharing also poses serious risks and ethical challenges that must be managed with care:
Privacy Breaches
Humanitarian datasets often contain personal details, such as names, locations, and family ties, about people who may already be at risk. If such data is leaked or accessed by malicious actors, it can expose people to further harm or exploitation. Exposing the identities or coordinates of aid recipients in a conflict zone, for example, could make them targets. Even aggregated, non-personal data can be sensitive in the wrong hands, such as maps of ethnic communities or shelter locations. Maintaining confidentiality and strong cybersecurity is literally a matter of safety in these contexts.
Lack of Consent and Community Trust
In crises, data is often collected under pressing conditions, and individuals may not fully understand or agree to how their information will be used. GiveDirectly's experience is instructive. When it interviewed 67 recipients in Malawi and Kenya in 2024 about using phone data to determine eligibility, almost all were comfortable with GiveDirectly itself accessing their data, though the organization cautions this may partly reflect a desire to take part. Recipients were far more worried about their data reaching family and community members, and neighbours learning about their poverty. GiveDirectly also found that its first, fact-based explanation of the AI approach was hard to understand, and rebuilt it around everyday scenarios (GiveDirectly). The lesson: consent is only meaningful if people understand what they are agreeing to, in their own language and terms.
Data Misuse and Security
Sharing data broadly can lead to unintended misuse if proper controls are not in place. If beneficiary data shared with a partner agency is later used for unrelated purposes, such as marketing, surveillance, or political targeting, it violates the humanitarian principle of do no harm. A notable controversy arose in 2019 when WFP announced a five-year, US$45 million partnership with the data analytics firm Palantir; privacy advocates questioned whether adequate safeguards were in place, while WFP stated that no data would be shared and personal data would remain under its control (The New Humanitarian). Sharing data with host governments can be equally sensitive: it may improve coordination, but authorities could also use data on refugees or minorities in ways that breach human rights.
Political Sensitivities
Humanitarian data often has political implications. Statistics on hunger, displacement, or casualties can embarrass governments or attract media scrutiny, and agencies sometimes face pressure to restrict or sanitize findings. Data sharing arrangements must allow critical information to flow for aid purposes while respecting laws and local contexts, often by publishing anonymized, aggregated figures rather than granular data. Balancing transparency with diplomacy requires careful judgment.
These challenges underscore the emergence of a culture of data responsibility in the humanitarian sector, defined in the IASC Operational Guidance as the safe, ethical and effective management of personal and non-personal data for operational response (OCHA Centre for Humanitarian Data). How can organizations achieve this in practice? The answer lies in robust governance.
A Framework for Responsible Data Governance
To share data responsibly, humanitarian organizations are developing comprehensive governance frameworks. Below is a practical, policy-driven framework covering the key pillars of responsible data sharing in humanitarian action.
1. Informed Consent and Community Engagement
Obtain informed consent before collecting or sharing personal data, unless doing so would compromise the urgency of life-saving measures. Explain why data is being collected and how it will be used, in a language and manner people understand, and test that explanation with them, as GiveDirectly did. Use culturally appropriate consent processes and allow people to opt out where possible. When individual consent cannot be obtained, for example during mass registration in a refugee influx, apply the do no harm principle, be transparent after the fact, and strictly limit use to the stated humanitarian purpose.
2. Data Minimization and Purpose Limitation
Collect and share only the data necessary for the task, and use it strictly for the defined humanitarian purpose. As a UNHCR officer working on the Tanzania data exchange put it: "We don't share more data than absolutely necessary; to minimize, as much as possible, the risks that come with regular data sharing" (UNHCR). Share names or biometrics only when required to deliver aid; otherwise use coded IDs or aggregates. GiveDirectly's practice when using phone data is a useful checklist: data protection impact assessments, data-sharing protocols in formal agreements, minimizing what is exchanged, pseudonymizing phone numbers, and informed consent (GiveDirectly). Purpose limitation builds trust that data will not creep into unintended uses.
3. Secure Access Controls and Anonymization
Not everyone needs access to raw personal data; most analysis can be done on anonymized or aggregated data. Use role-based access controls, encryption, and audit trails so only authorized personnel and partners can reach sensitive data. HDX embodies a tiered approach: data is shared publicly, privately within an organization, or through HDX Connect on approved request, depending on its sensitivity, and the most sensitive data should only be shared with strict security measures and a data sharing agreement where required (HDX documentation). Regular security audits and mandatory staff training maintain confidentiality and integrity.
4. Data Sharing Agreements and Legal Compliance
Formalize data sharing through clear agreements that specify what data will be shared, for what purpose, who will access it, how it will be protected, and how long it will be retained. These agreements should align with international data protection principles, such as the UN Personal Data Protection and Privacy Principles, and any applicable law. A leading example is the 2018 Addendum on Data Sharing to the WFP–UNHCR global memorandum of understanding, which grounds all data exchanges between the two agencies in international and UN data protection and privacy standards. By codifying expectations, agreements create accountability.
5. Oversight, Accountability and Continuous Improvement
Establish governance bodies and processes to oversee data activities: a data protection officer or data governance committee, regular audits and risk assessments, and an incident response plan. At the level of a whole response, the most widely adopted data responsibility action is the Information Sharing Protocol, which serves as the primary reference document for data and information sharing in a response (OCHA Centre for Humanitarian Data). When mistakes or breaches occur, organizations must be transparent and remediate quickly; this accountability strengthens trust over time.
From Policy to Practice: Roles in Operationalizing Data Governance
Translating policies into day-to-day practice requires dedicated roles and skills. Data analysts, data scientists and IT specialists implement the principles: anonymizing datasets before sharing, running disclosure risk checks on outputs, and building systems that are secure by design. Data governance and protection specialists develop policies, conduct training, and provide oversight.
Collaboration between technical and policy staff is key, because a framework only matters if front-line staff understand and embrace it. That is why leading organizations invest in capacity building: GiveDirectly, for example, reports that 93% of its more than 800 staff completed its eight mandatory risk training modules in 2025 (GiveDirectly). The message is that everyone handling data is a steward of it and must uphold the organization's values when using it.
Championing Responsible Data for Humanitarian Impact
As the humanitarian sector becomes increasingly data-driven, responsible data governance is not a box-ticking exercise; it is fundamental to principled and effective action. Sharing data responsibly enables life-saving coordination, transparency, and innovation while guarding against harm to the very people we aim to help.
From large UN agencies such as WFP, UNHCR and OCHA to agile innovators such as GiveDirectly, organizations are converging on the same truth: without robust policies on consent, privacy, and accountability, the promise of data can too easily turn into peril. When data is governed well, it becomes a powerful enabler that amplifies impact and upholds human dignity. The challenge now is to scale these practices across all operations and organizations.
Looking Ahead: AI, Innovation, and My Perspective
The next frontier for humanitarian data governance is not only about sharing, but about how emerging technologies use what is shared. Machine learning models are already being trained on humanitarian datasets. This offers enormous promise, from forecasting famines to optimizing cash transfers and pre-positioning supplies ahead of disasters, but it also raises critical questions: who owns the training data, what safeguards protect against bias or re-identification, and how do we ensure algorithmic decisions remain accountable to the communities they affect?
This is where the discipline of data governance becomes transformative. Humanitarian organizations can lead by adopting practices such as federated learning, in which models are trained across multiple agencies' data without moving the data itself, or differential privacy, which enables insights without exposing individuals. These are not futuristic luxuries; they are tools we can deploy now to make data sharing both useful and safe.
Conclusion
I see a dual mission: to strengthen the integrity of humanitarian data pipelines today and to prepare organizations for the ethical use of tomorrow's technologies. Robust frameworks, clear consent, and strong accountability mechanisms are not obstacles but enablers, the foundation that allows innovation and trust to move together.
If your organization is grappling with these questions, I would welcome the chance to exchange ideas. The humanitarian sector cannot afford to treat data governance as an afterthought; it must become a core capability. Done right, data sharing can be not just an operational practice, but a pathway to more effective, equitable, and resilient humanitarian action.
