These publishers are discovery targets. Inclusion does not mean every article is false. Each factual claim is evaluated independently against original evidence.
35.4% problematic-claim rate across 82 evaluated claims.
ranked sample · rank #4
46 published investigations
Visit publisher ↗ View our investigations →No scored claims yet
0 published investigations
Visit publisher ↗ View our investigations →29.5% problematic-claim rate across 44 evaluated claims.
ranked sample · rank #5
26 published investigations
Visit publisher ↗ View our investigations →40.0% problematic-claim rate across 5 evaluated claims.
building sample
4 published investigations
Visit publisher ↗ View our investigations →39.4% problematic-claim rate across 104 evaluated claims.
ranked sample · rank #1
51 published investigations
Visit publisher ↗ View our investigations →38.9% problematic-claim rate across 54 evaluated claims.
ranked sample · rank #2
34 published investigations
Visit publisher ↗ View our investigations →No scored claims yet
0 published investigations
Visit publisher ↗ View our investigations →36.5% problematic-claim rate across 85 evaluated claims.
ranked sample · rank #3
56 published investigations
Visit publisher ↗ View our investigations →28.6% problematic-claim rate across 56 evaluated claims.
ranked sample · rank #7
31 published investigations
Visit publisher ↗ View our investigations →No scored claims yet
0 published investigations
Visit publisher ↗ View our investigations →50.0% problematic-claim rate across 6 evaluated claims.
building sample
5 published investigations
Visit publisher ↗ View our investigations →23.1% problematic-claim rate across 26 evaluated claims.
ranked sample · rank #6
16 published investigations
Visit publisher ↗ View our investigations →The collector checks official RSS feeds first. It also discovers alternate RSS or Atom links advertised by the publisher, parses political section pages, and uses the publisher’s own sitemaps when feeds are unavailable or incomplete.
The score is not the number of articles and not the number of mistakes. It combines the average severity of evaluated claims with the rate of evaluated stories containing a documented problem. A publisher that releases more articles receives no automatic penalty.
Small samples are shrunk toward the sitewide average. A score is marked provisional only after at least 8 evaluated claims across 4 stories, and fully ranked after at least 20 claims across 8 stories. Opinions, hyperbole and inconclusive verdicts are excluded from the calculation.
Monitored outlets supply claims to examine. They are not automatically treated as evidence. Verdicts prioritize primary records, official data, court documents, full transcripts and original research.