Trust
HAPPND organises what happened, not publishers. The persistent object is the Story: a container that holds every development, and every article, video, document or official statement attached as evidence.
HAPPND ingests publicly available metadata: publisher and agency feeds, official government and institutional sources, and video platforms via their official APIs and embeds. We store headlines, descriptions, permitted snippets, timestamps, language, location and provenance — never full publisher article bodies, and never downloaded video files.
Incoming content is normalised, de-duplicated, language-tagged and geolocated, then matched to an existing Story or proposed as a new one. Notable changes become developments on the Story timeline. Ranking weighs velocity, geographic spread, source diversity, primary-source activity and persistence — coverage volume alone is not importance.
AI assists with clustering, summarising, entity extraction and research packaging. It does not decide what is true. HAPPND does not label publishers ideologically and does not present itself as an arbiter of fact. Where a summary exists, the underlying evidence is always listed beneath it with a link to the original.
Automated organisation makes mistakes: stories can be merged wrongly, a development can be missed, a translation can mislead, and a summary can flatten nuance. Live research depends on third-party availability. Treat HAPPND as a map of the reporting, and read the originals before you rely on anything.
Spot something wrong? Tell us on the Sources & corrections page and we will review it.