Who Owns Disaster Data?
Ethics in Post-Crisis Environments
When disaster strikes, information becomes one of the most valuable resources available. Satellite imagery is captured within hours, drones survey damaged infrastructure, emergency responders collect photographs from the ground, and local communities contribute reports through mobile devices and social media. Together, these diverse sources of information help create a rapidly evolving picture of events, enabling authorities to prioritise rescue efforts, assess damage and begin planning recovery.
Artificial intelligence has significantly increased our ability to process this information at scale. Machine learning models can identify damaged buildings, estimate flood extents, detect blocked transport routes and combine multiple sources of evidence into actionable insights far more quickly than manual analysis alone.
Yet alongside these technological advances lies a more complex question.
Who owns the data generated during a disaster?
Unlike many commercial datasets, disaster data often originates from communities experiencing extraordinary circumstances. It may contain sensitive information about homes, infrastructure, cultural heritage, businesses and individuals at their most vulnerable. It may cross national borders, involve multiple agencies and be shared internationally in the interests of humanitarian response.
The technical challenges of analysing disaster data are increasingly well understood. The ethical challenges surrounding its ownership, governance and long-term use are only beginning to receive the attention they deserve.
As AI becomes more deeply integrated into disaster resilience, establishing clear principles for data stewardship will be just as important as improving model performance.
Data Created During a Crisis Is Different
Many discussions around AI governance focus on familiar questions of privacy, intellectual property or commercial licensing. Disaster data presents a different set of considerations.
Information collected during an emergency is rarely gathered for commercial purposes. Its primary objective is to support life-saving decisions, coordinate response efforts and reduce harm. This urgency often necessitates rapid sharing between organisations that may not ordinarily exchange data.
At the same time, the information being collected can be highly sensitive. High-resolution imagery may reveal the condition of private homes, critical infrastructure or culturally significant sites. Geospatial datasets can expose vulnerabilities that persist long after the immediate emergency has passed. Personal reports submitted by affected communities may include information that individuals never expected to become part of long-term datasets.
Unlike conventional data collection, consent is often impractical during an unfolding crisis. Communities affected by disasters are understandably focused on immediate safety rather than considering how their information may be stored, analysed or reused in the future.
This creates a responsibility for organisations developing AI systems to adopt ethical governance from the outset, rather than relying solely on legal compliance.
Ownership, Stewardship and Responsibility
It is tempting to think of data ownership in purely legal terms, but disaster resilience requires a broader perspective.
Rather than asking who owns the data, we might instead ask who is responsible for looking after it.
Responsible stewardship encompasses far more than secure storage. It includes:
understanding where data originated;
maintaining accurate records of provenance;
documenting how datasets have been processed;
ensuring appropriate access controls;
respecting cultural and national sensitivities;
considering how data may be reused long after the immediate crisis has ended.
This is particularly important where international collaboration is involved.
Major disasters often bring together governments, humanitarian organisations, research institutions and private sector partners from multiple countries. While this collaboration enables more effective response, it also introduces differences in legal frameworks, data governance standards and expectations around ownership.
Without clear agreements, valuable information can become fragmented or difficult to access precisely when it is needed most.
Good governance therefore becomes an enabler of collaboration rather than an administrative burden.
Data Sovereignty in an Interconnected World
Disasters do not respect national borders, but data governance frequently does.
Many countries are placing increasing emphasis on data sovereignty: the principle that information generated within a nation should remain subject to its own legal and regulatory frameworks. This is particularly relevant for datasets relating to critical infrastructure, public services or national security.
At the same time, international cooperation is often essential during major emergencies. Satellite imagery may originate from one country, AI models may be developed in another, while humanitarian organisations coordinate response across multiple jurisdictions.
Balancing these competing priorities requires careful planning.
Rather than assuming unrestricted data sharing, organisations should design systems that support interoperability while respecting national governance requirements. Approaches such as federated learning, secure data environments and privacy-preserving analytics may help enable collaboration without requiring unrestricted movement of sensitive information.
Technology alone, however, cannot solve governance challenges. Trust between institutions remains fundamental.
Transparency Builds Public Trust
Communities affected by disasters are more likely to support the use of AI when they understand how their information is being used.
Transparency should therefore extend beyond explaining AI models. It should also encompass the data itself.
Organisations should be able to answer straightforward questions:
What data has been collected?
Why was it collected?
Who can access it?
How long will it be retained?
How is it protected?
Can it be reused for research or future preparedness?
Under what conditions will it be deleted?
Providing clear answers to these questions helps establish confidence that technology is serving the interests of affected communities rather than exploiting their circumstances.
Trust, once lost, is difficult to rebuild.
Ethical Data Enables Better AI
Good governance is sometimes viewed as slowing innovation. In practice, the opposite is often true.
AI systems are only as reliable as the data on which they are trained and evaluated. Understanding provenance, documenting collection methods and maintaining high-quality metadata all contribute directly to more robust models.
Ethical data management also supports explainability. When decision-makers understand where information originated and how it has been processed, they are better placed to evaluate the recommendations generated by AI systems.
For SMEs developing innovative technologies, this presents an important opportunity. Building transparent governance into products from the beginning can become a competitive advantage as customers increasingly seek solutions that are both technically capable and demonstrably trustworthy.
Similarly, public sector organisations and heritage bodies managing sensitive datasets will increasingly expect technology partners to demonstrate responsible stewardship rather than simply technical capability.
Final Thought
Disaster data is more than a technical resource. It represents communities, places and lives affected by extraordinary circumstances.
As AI becomes an increasingly important component of disaster resilience, conversations about model performance must be accompanied by equally serious discussions about governance, stewardship and responsibility.
The question is not simply who owns disaster data. It is whether we can build systems that respect the people behind the data while enabling the collaboration needed to protect them.
Responsible innovation begins long before an algorithm is deployed. It begins with the principles that govern the information on which that algorithm depends.
