Editorial standards and AI policy
Updated April 2026. We publish humanitarian data analysis for researchers, journalists, and policymakers. The standards below describe how we source, review, and correct everything on the site.
Our sources
Every figure we publish traces back to a primary humanitarian data source. We do not invent statistics, and we do not cite anonymous social posts as evidence. Our core source set:
- • UNHCR Refugee Data Finder and country operational portals
- • OCHA Humanitarian Data Exchange (HDX) and Humanitarian Needs Overviews
- • IOM Displacement Tracking Matrix (DTM)
- • ACLED for conflict event data
- • IDMC Global Internal Displacement Database
- • ReliefWeb for situation reports and humanitarian appeals
- • REACH Initiative for needs assessments
- • World Bank and UNDP for socioeconomic indicators
How we use AI
We use large language models to draft synthesis paragraphs and daily briefings from current reporting. Every AI-assisted page is clearly labeled with an amber disclosure banner and meets four conditions:
- The model is given only verified primary-source documents as context.
- Each draft is checked against its sources before publication.
- Sources are cited inline at the foot of the page so readers can verify every figure.
- A site-wide content sanitizer strips filler language and AI-typical phrases.
We do not use AI to generate quotations, eyewitness accounts, or images. We do not use AI to invent statistics or to bypass paywalls.
Who is behind the platform
Humanity Centered Data is an independent open research platform. We have no affiliation with any UN agency, government, NGO, or political party. We accept no payment for placement and we run no sponsored content. The site is funded by contextual display advertising and by readers who use the embeddable widgets and the public API.
Corrections policy
We correct errors of fact within 24 hours of being notified. Substantive corrections are noted at the foot of the affected page with the date the correction was made.
To request a correction, write to our editors with a link to the page, the figure or sentence in question, and the primary source you believe to be correct.
What we will never do
- • Publish anonymous, unverifiable, or fabricated statistics.
- • Take payment to alter, remove, or feature analysis.
- • Run advertising next to graphic descriptions of violence or images of victims.
- • Republish copyrighted reports without permission.
- • Endorse a political party, government, or armed group.