How this works

We pull headlines and short excerpts from outlet RSS feeds — never full articles. Each item is turned into a numeric embedding, and items about the same event are grouped into a story by similarity within a rolling time window. A story reaches the feed once it spans at least two sides of the spectrum.

Each outlet's left / center / right label comes from published, independent media-bias methodologies — AllSides and Ad Fontes Media. They are not our own judgment. Reasonable people disagree about any single label; we cite the source so you can check it.

For each multi-side story, a large language model reads the collected headlines and excerpts and produces three things: the shared facts every side reports, a per-side note on how the language differs, and the specific loaded termseach side used (highlighted in that side's color).

This analysis is AI-generated and clearly labeled as such. It describes how coverage differs — it is not a ruling on which side is correct, and it can be wrong. Treat it as a reading aid, not a verdict.

We store and show only headlines, a short excerpt, and a link back to the original article. Every headline links out to the outlet that published it — read the source.

The feed currently draws from these outlets:

The source list grows over time; lean labels are reviewed against AllSides / Ad Fontes as ratings change.