One story. Compare the coverage.

What is measured: an AI model assesses the political framing of each headline and up to 240 characters of its publisher-provided RSS excerpt. No predefined publisher-lean ratings are supplied to the model. Full articles, images, and paywalled text are not read. The result is a provisional estimate of this short text, not a verified rating of an entire outlet or an assessment of factual accuracy.

Scale: −100 is strongly left-framed, +100 strongly right-framed. Scores below −20 are Left, scores from −20 through +20 are Center, and scores above +20 are Right. Center means the available text was assessed as descriptive or balanced. The model must return an explanation and exact quoted evidence; results without valid evidence or with self-reported confidence below 55% remain Uncertain. Confidence is the model's self-assessment, not a calibrated statistical probability.

Counts and percentages: reports are grouped by publisher, with equal weight per publisher. Scores from its assessed reports are averaged. A publisher with both Left and Right reports is marked Mixed instead of being averaged into Center. The L / C / R bar uses only those three classified categories as its denominator; Mixed, Uncertain, and Pending counts remain visible separately. Partial assessments are identified in the source list. The overall score averages assessed publishers equally; a single-source estimate is clearly marked.

Coverage analysis by Ask Toaster

Ask Toaster powers our local AI assessments of headlines and feed excerpts, with no paid API. Validated results are published and cached. New reports are checked every 15 minutes while the local computer and its AI service are running. Selecting an unanalyzed story moves it into the next run’s priority queue; results stay pending until an assessment is available. Each result retains the model, analysis time, and text used for the assessment.

Coverage limits: article counts include only collected feeds. Similar headlines within a 72-hour window suggest related stories; automated matches can be wrong. Publisher locations describe the source, not the event location. Last updated reflects the feed check; article publication times are shown separately. Multiple feeds from CNBC, The Guardian, and NOAA each count once within their publisher group.

Independent methodology: this is The Meyers Report's AI-assisted content assessment. Ground News uses publisher-level bias ratings; these are not Ground's ratings. AI may misread quotations, satire, context, or political framing. Read the evidence and original reports before relying on a result.