GreyMatter is the analysis engine behind GreyNews. It reads news articles, extracts factual claims, cross-checks them against authoritative data sources when possible, detects framing and language patterns, and synthesizes how multiple outlets covered the same story.
If you want the scoring mechanics behind the feed, read the methodology page.
GreyMatter uses a routed analysis pipeline: extraction, evidence matching, framing checks, clustering, ranking, and synthesis each run as separate steps with versioned instructions and reviewable outputs. The public product explains what each step is allowed to conclude without exposing vendor internals or implementation details that would make the system easier to game.
GreyNews is designed to keep incentives legible: source links stay visible, uncertainty is labeled, and generated summaries are constrained by evidence from the underlying article set. We publish corrections and methodology notes so readers can challenge the work without needing access to internal infrastructure.
GreyMatter describes what sources report and what the cited evidence supports. Public summaries should frame claims as supported, disputed, or unverifiable rather than declaring the world settled.
If you spot a mistake, report it to corrections@greymatter.news. Corrections are logged publicly.
The companion explanation of scoring and verification lives on the methodology page.