Open news datasets, 2019–2026

Sixty topic datasets drawn from the Prism news index, a global index of news articles, each covering January 2019 – July 2026. Every release has an article-level CSV sample, monthly counts, coverage by country and by US and Indian state, tone and themes, a research paper, and findings written for citation. Together they count 17,376,896 articles.

• Free to use under CC BY-NC 4.0, with attribution and a link.
• Mirrored on Hugging Face, Kaggle, Zenodo (with DOIs), the Internet Archive and Harvard Dataverse.
• Reproducible: every number comes from the Prism News API, which you can query yourself.