What You See for Each Article
For every article, Meridian produces a set of AI-generated observations designed to help you read more critically. These include: a factual summary of what happened (stripped of framing and opinion), whose perspective the story is told from, specific language choices that shape interpretation, assumptions the article takes for granted, perspectives or voices that are absent from the coverage, and the overall tone. None of these are scores or verdicts — they are observations for you to weigh against the original article.
How the Analysis Works
Each article goes through an automated analysis pipeline. First, verifiable facts are extracted — who, what, when, key outcomes — separated from interpretation. Then a second pass generates observations about framing: whose perspective dominates, what language choices were made and what they cost the reader, what assumptions go unchallenged, and which angles are structurally absent. The goal is surfacing patterns that are easy to miss on a single read-through, not rendering judgment on the journalism.
Story Clusters
When multiple sources cover the same story within a 96-hour window, Meridian groups them into a cluster. Comparing how different outlets frame the same event — which facts they foreground, whose voice they centre, what they omit — is one of the most reliable ways to understand what's actually happening. Cluster pages show the same story through different editorial lenses.
Coverage Gaps
Weekly analysis compares which topics and perspectives are systematically absent from the sources Meridian monitors. A gap is flagged when a significant story receives coverage from only one outlet, or when an affected community's perspective is absent from all coverage. This analysis reflects the limitations of our source set — we don't monitor all publishers.
Source Selection
Meridian currently monitors RSS feeds from: The New York Times, NPR, BBC, The Guardian, The Wall Street Journal, The Economist, Al Jazeera English, Deutsche Welle, ProPublica, The Intercept, and Axios. Sources were chosen to represent a range of editorial perspectives, geographic vantage points, and institutional structures. NYT and WSJ articles are often stubs due to paywall restrictions and may show reduced analysis quality.
Limitations
Meridian's analysis is fully automated using AI. AI analysis can be wrong — it may misread tone, miss context, or identify patterns that aren't meaningful. Observations are heuristic, not ground truth. The system reflects the biases present in its training data and the choices made in its prompts. All observations should be verified against the original article before drawing conclusions. We publish this methodology openly so users can critique it.
Meridian is an independent tool built to support media literacy. It is not affiliated with any news organisation. Analysis is AI-assisted and provided for educational purposes only — not as a substitute for professional journalism, legal advice, or editorial fact-checking.