Two of the most-cited media bias charts refreshed within weeks of each other. AllSides shipped Version 11 of its chart in December 2025, and Ad Fontes released its first 2026 edition. Both are now circulating, and a lot of people are trying to work out which one to believe. Here is a fair look at how each is built, where they line up, where they don't, and the one limitation neither chart can escape.
How AllSides builds its chart
AllSides rates bias, not reliability. Its Version 11 chart places outlets on a single left-to-right scale — Left, Lean Left, Center, Lean Right, Right — using two methods that are deliberately kept separate. The first is the Blind Bias Survey: everyday Americans across the political spectrum rate news content without being told which outlet produced it, which strips out the halo effect of a familiar logo. The second is an Editorial Review by a multipartisan panel trained to spot bias. When the two disagree, AllSides makes a judgment call.
Version 11 added new outlet reviews, retired some old branding, took a harder look at disputed sources like The Epoch Times, and folded in several newsletters. Reuters was re-confirmed as Center. The newsletters 1440, Tangle, and Morning Brew all landed at Center. The Atlantic sits at Left, TheBlaze at Right, ZeroHedge at Lean Right.
The most revealing entry is the Wall Street Journal's opinion section. In the August 2025 blind survey, 543 respondents rated WSJ Opinion Lean Left on average. A multipartisan editorial panel, reviewing the same section three months later, rated it Lean Right. AllSides kept the Lean Right label. Two rigorous methods, one outlet, opposite answers — a useful reminder that even the chart's makers know a single label is a compression.
How Ad Fontes builds its chart
Ad Fontes rates two things at once. Its chart is a grid: a horizontal axis for political bias (left to right) and a vertical axis for reliability, running from original fact reporting at the top down through analysis, opinion, and, at the bottom, fabricated information. Reliability uses a 0–64 scale, where scores above 40 are generally solid and below 24 are generally trouble.
The ratings come from a team of more than 40 analysts. Each article or episode is scored by a pod of three — one who identifies as left, one center, one right — and the scores are averaged. Bias breaks into Political Position, Language, and Comparison; reliability into Expression, Veracity, and Headline/Graphic. The 2026 edition leans on more granularity and adds AI assistance to speed up content evaluation, along with more podcasts and independent newsletters than earlier versions.
The practical difference: AllSides tells you which direction an outlet leans. Ad Fontes tells you that and how far you can trust the reporting underneath it. If you only care about lean, AllSides is simpler. If you want a trust axis too, Ad Fontes gives you one.
Where they agree — and where they don't
On the big, obvious cases, the two charts mostly rhyme. Wire services cluster near the center, opinion-heavy cable and partisan sites sit out at the edges, and almost nobody argues about the extremes. The disagreements show up in the middle, where methodology matters most: an outlet AllSides calls Center might land slightly left or right of the Ad Fontes centerline, because one chart is measuring perceived bias and the other is scoring sampled articles.
This is also where a common question lands: is CNN Lean Left or Left? The honest answer is that it depends on what you're measuring — the news desk and the opinion programming don't rate the same, and a chart that publishes one dot per outlet has to average across both. Neither chart is wrong. They're answering slightly different questions with slightly different inputs.
The limitation neither chart escapes
Both charts share one structural constraint: they plot an outlet as a single point. That point is an average across months of output. It cannot tell you how the outlet framed the specific story you read this morning — and framing is where most of the day-to-day slant actually lives.
Here's a concrete example from our own data this week. When Trump used a NATO summit to announce the US would let Ukraine manufacture Patriot missiles, The New York Times led its live coverage with "Trump Lashes Out at Europe at NATO Summit," folding the Patriot decision into a separate, quieter piece. The Washington Examiner led directly with "Trump says US will grant Ukraine license to produce Patriot Missiles," and The Hill specified "co-produce key air-defense weapons" in its first sentence. Same event, same facts, two entirely different choices about what the headline should be. No dot on any chart captures that.
The pattern repeats. On the morning Trump declared the Iran ceasefire "over," the NYT's live blog framed it as "Trump Casts Doubt on Future of Cease-Fire After Latest Strikes," while Bloomberg's headline was "Stocks Drop, Oil Jumps After Trump Says Ceasefire with Iran Is 'Over'." One outlet led with diplomacy, the other with markets — 29 sources covered the same story, and the framing split cleanly along what each newsroom decided mattered. A bias chart would tell you both outlets sit near the center. It would not tell you they told you to worry about different things.
This is the gap we've written about before in what a media bias chart can't tell you and why media bias ratings aren't enough: a rating is a starting point, not the whole story.
So which one should you trust?
Both, for what each is good at. Use AllSides when you want a fast read on an outlet's general orientation. Use Ad Fontes when you also want a reliability check on the reporting. Neither is a scam, and the fact that they occasionally disagree is a feature — it tells you the outlet lives in contested territory rather than settled ground.
But treat the chart as the first click, not the last. An outlet's label is fixed; its framing changes story to story. The only way to see that is to read the same event across the spectrum and watch what each newsroom leads with, buries, or leaves out. That's the layer charts can't reach, and it's the one that changes what you actually walk away believing. If you're assembling a broader toolkit, our roundup of the best media bias tools in 2026 covers where each of these fits.