Explainer
What Is Media Bias?
Media bias is the slant a news source introduces through word choice, story selection, and framing. It does not require malicious intent — it can be structural, ideological, or simply the result of who is in the newsroom. DailyComposite uses four AI models to measure it on every article it indexes.
The Four Types
How Bias Shows Up in News
Selection Bias
Which stories get covered at all. A news organization that consistently covers one side's scandals while ignoring the other's is exercising selection bias, regardless of whether each individual story is reported accurately.
Framing Bias
How a story is presented. Two outlets can report the same event with very different word choices, headline angles, and source selection, leading readers to opposite conclusions from the same underlying facts.
Omission Bias
What gets left out. A story can be technically accurate yet deeply misleading if it omits key context, counterarguments, or inconvenient data. Bias by omission is often harder to detect than outright spin.
Spin
The addition of editorial interpretation to factual reporting. Spin appears as loaded adjectives, emotional language, unnamed sources used to advance a narrative, or conclusions that go beyond what the facts support.
Why It Matters
The Real Cost of Unchecked Bias
Most people read news from a narrow set of sources. If those sources share a consistent political lean, readers can form an accurate picture of one side of a debate while having almost no exposure to the other. That is not a personal failure. It is a structural one.
Research on media consumption shows that exposure to a single-perspective news diet correlates with more polarized political views over time. This is not because readers are gullible. It is because framing shapes what questions feel worth asking.
Measuring bias is not the same as correcting for it. But making it visible gives readers the ability to account for it. A score of 2 on a story does not mean you should ignore it. It means you should find the version of the same story that scores a 4, and read both.
Our Approach
How DailyComposite Measures Media Bias
DailyComposite does not rely on human editorial judgment or hand-curated outlet labels. Instead, every article is sent to four large language models — Claude, GPT-4o, Grok, and Gemini — which read the article text and independently score it on a 1-5 scale.
A score of 1 is strongly left-leaning. A score of 5 is strongly right-leaning. A score of 3 is neutral. The four scores are averaged into a single consensus number that appears alongside every article on the site.
Using four independent models reduces the risk that any single model's training-data biases dominate the result. When all four models agree, the consensus is strong. When they diverge, the spread itself tells you something about the article.
FAQ
Common Questions
Is media bias the same as fake news?
No. Bias describes a slant in how real events are framed or selected. Fake news refers to fabricated stories. A biased outlet can report true events while still presenting them in a misleading way.
Are all news outlets biased?
To some degree, yes. Every editorial decision — what to cover, how to headline it, which sources to quote — reflects a set of values. The question is not whether bias exists but how pronounced it is and in which direction.
Can I trust an outlet with a bias score of 2 or 4?
Bias is not the same as unreliability. A leaning outlet can report accurately. The score tells you about framing, not credibility. Use it to read more diversely, not to dismiss sources.
How often are scores updated?
DailyComposite scores articles at the time of ingestion. Scores are not retroactively updated unless an article is re-analyzed through the Bias Checker tool.
Who built DailyComposite?
DailyComposite is an independent project. It does not have editorial relationships with any news outlet. Learn more on the About page.
See bias scores in real time
Browse today's top stories with bias scores on every headline. Back to homepage.