Project Guardian was developed to help communities, nonprofit organizations, public agencies, and practitioners better understand patterns of housing instability, heirs' property vulnerability, displacement risk, and related community conditions. The platform is intended to support informed decision-making, responsible intervention, and equitable resource allocation, not exploitation, surveillance, or harm.
Because data-driven tools carry important ethical responsibilities, the Project Guardian team and their partners believe it is essential to establish a shared understanding of how information should be interpreted, discussed, and applied in practice. The principles below are intended to serve as a practical guide for responsible stewardship of data and insights generated through Project Guardian and related collaborative efforts.
These principles are not presented as a legal agreement or binding policy. Rather, they reflect a shared commitment to thoughtful, ethical, and community-centered use of information. By using Project Guardian, users acknowledge the importance of these principles and the responsibility that comes with working with sensitive housing and community data.
Project Guardian recognizes that data alone cannot fully represent the lived experiences of individuals, families, or neighborhoods. Outputs should therefore always be considered alongside local knowledge, community context, and professional judgment.
Use data to protect, not to exploit
Every analysis should contribute to keeping people housed, informed, and supported. If a use case creates risk of harm, it does not proceed.
Start with people, not parcels
Data may be organized at the parcel level, but decisions consider the human context: households, histories, and community conditions.
Minimize harm through restraint
Not all insights need to be exposed. Sensitive signals are shared only when there is a clear, constructive use and an appropriate audience.
Be precise about uncertainty
Outputs reflect ranges, likelihoods, and limitations. Avoid overstating confidence, especially when results may influence funding, policy, or intervention.
Prioritize transparency in method, not exposure of individuals
Methodologies, assumptions, and logic remain open and explainable. Individual-level data stays protected and controlled.
Design for action, not surveillance
The platform exists to guide decisions and resource allocation, not to monitor or track individuals without purpose or consent.
Respect ownership and stewardship of data
Data sources, especially those tied to communities, are handled with care. Use aligns with original intent, legal requirements, and ethical responsibility.
Avoid reinforcing structural inequities
Models are evaluated for bias and unintended consequences. Outputs are reviewed in context to ensure they do not amplify existing disparities.
Share insights with those who can act, and those affected
Decision-makers receive clear, usable outputs. Communities are not excluded from understanding patterns that affect them.
Stay accountable and adaptable
Assumptions are revisited. Feedback is incorporated. When something does not hold up in practice, it is corrected.