Nonprofits work with people who trust them with sensitive data, personal stories, and sometimes their safety. Adding AI does not erase these duties. It raises the stakes. The old rules that guide ethical work still apply, and several NGOs show how to use them when they add AI tools to daily operations.
The first rule is data protection. GDPR at https://gdpr.eu sets a clear baseline. You collect only what you need, you disclose why you collect it, and you store it securely. You also avoid passing personal data to systems you cannot control. The Norwegian Refugee Council at https://www.nrc.no follows these steps when it tests language models for field reports. The team strips personal details, keeps tools offline, and requires staff checks before a report leaves the system. This protects staff, clients, and partners from exposure.
The second rule is informed consent. People have the right to understand how their data is used. HomeLink Outreach in Canada at https://homelinkoutreach.ca uses a simple classifier from Hugging Face at https://huggingface.co/models to help sort emails. They train the model with anonymized data. They also post a short public notice on their website that explains how the system works, what it does not do, and who reviews the final decisions. This aligns with guidance from the Canadian Office of the Privacy Commissioner at https://www.priv.gc.ca.
The third rule is accountability. You stay responsible for decisions, even when AI helps. Médecins Sans Frontières at https://www.msf.org uses a closed model to summarize clinical guidance. Staff rely on these summaries to save time, but clinical judgment remains with medical teams. Their approach reflects the World Health Organization’s data protection standards at https://www.who.int/publications/m/item/policy-on-data-protection. AI supports decisions but never replaces them.
The fourth rule is transparency. People deserve to know what tools shape reports, ratings, or outreach messages. Charity Navigator at https://www.charitynavigator.org uses AI to sort financial data. They follow documentation from the US Federal Trade Commission at https://www.ftc.gov/business-guidance. They publish a clear list of data sources and update it when methods change. This helps donors understand how ratings work and keeps the system credible.
These rules show up across sectors.
• Use only the data you need.
• Remove personal identifiers before training any model.
• Explain how AI tools shape services or messages.
• Keep humans in charge where it affects rights or safety.
• Follow public standards from GDPR, WHO, the FTC, and national privacy authorities.
AI can save time and improve outreach. It cannot replace the duties that protect people’s rights and privacy. The old rules still hold. They keep nonprofits accountable and maintain the trust that makes their work possible.
