Our Standards
The rules we write by. In public.
These are the rules every story is written to, and the checklist every story must pass before anything is published.
01Our promise
Report what can be verified. Show the disagreement.
We explain what happened, what is verified, what is disputed, and how outlets across the political spectrum are framing it. Where people disagree, we show you the disagreement instead of settling it for you.
We cover seven topics: the U.S., the world, the economy, technology, science, health and culture. Culture means the arts, media, education, religion and the institutions behind them. We also cover major stories where every side agrees, because it's worth seeing that the facts aren't always up for debate. We leave sports scores, celebrity gossip, entertainment reviews and lifestyle news to others, unless a story has real public consequence.
If you find an article that breaks any rule on this page, tell us. That's a correction we want to make.
02Our sources
Where our stories come from.
We watch a pool of 34 news outlets: 13 on the left, 9 in the center and 12 on the right. Every one is rated by AllSides, reports its own news, and is free for you to read, so you can check any outlet we link. A story's priority depends partly on what share of each group covers it, so a longer list on one side never gives that side a bigger say.
Left
- CNNLean Left
- NBC NewsLean Left
- The GuardianLeft
- HuffPostLeft
- VoxLeft
- Associated PressLean Left
- CBS NewsLean Left
- ABC NewsLean Left
- PoliticoLean Left
- NPRLean Left
- AxiosLean Left
- PBS NewsHourLean Left
- USA TodayLean Left
Center
- ReutersCenter
- The HillCenter
- BBC NewsCenter
- The Christian Science MonitorCenter
- NewsweekCenter
- UPICenter
- NewsNationCenter
- Deseret NewsCenter
- ReasonCenter*
Right
- Fox NewsRight
- New York PostRight
- The Daily WireRight
- The Washington TimesLean Right
- The Daily CallerRight
- Just The NewsLean Right
- The Center SquareLean Right
- Washington Free BeaconRight
- NewsmaxRight
- The New York SunRight
- The Daily SignalRight
- The National DeskRight
Ratings from AllSides, checked November 10, 12, 13 and 16, 2026. *Reason's overall rating is under review. A Center rating doesn't mean more accurate, and a Left or Right rating doesn't mean less accurate. It describes perspective. We leave out outlets you'd have to pay to read, such as The Wall Street Journal and The New York Times.
How outlets are rated
We don't rate the outlets ourselves.
We place each outlet using the media bias ratings published by AllSides, an independent organization that rates news sources from Left to Right. AllSides combines several methods, including two that matter most here:
- Editorial reviews. A panel of reviewers from the left, center and right reads an outlet's coverage, looks for slant, spin and story choice, and agrees on a rating.
- Blind bias surveys. Everyday Americans from across the political spectrum rate an outlet's headlines and articles without knowing which outlet they're reading.
We recheck every rating every three months. When an outlet's rating moves out of its group, we move it to the group that fits. We also drop outlets that put up a paywall or stop reporting their own news.
Choosing stories
What people are talking about, weighed against what matters.
A story's priority comes from two scores.
Who's covering it
What share of each group's outlets carry the story, how prominently, whether all three groups cover it, and whether coverage is growing.
How much it matters
How many people it reaches, how lasting or hard to reverse the consequences are, and how much is genuinely new and confirmed.
Attention alone would miss important stories that happen quietly. Importance alone would miss what the country is actually talking about. We use both.
When one group's outlets barely cover a story the others are covering, a reader who follows only that side would likely miss it. We say the story was Out of Frame for that side's readers. We measure each group as a share of its own outlets, so groups of different sizes compare fairly. A story is Out of Frame for a group when none of its outlets has reported it, and at the Edge of Frame when that group is far behind the group covering it most. We give a story time to be picked up before we flag it, we only flag a gap that holds up across separate checks, and we search the quiet outlets ourselves to confirm they haven't covered it. The article says so plainly, with the counts and the time we took them, and still gives the quiet side's own view. If a gap is confirmed within two days after we publish, we add the label with a dated update. We report what each side's readers were shown, never why.
- Every side gets at least two voices. Each article draws on at least two outlets from each group in this list, and we go to primary sources (court filings, official data, transcripts) wherever we can. We may cite other outlets for a fact, but they never stand in for a side. Every source we used is listed at the end of each article.
- Challenges can cite anyone credible. If you think we got something wrong, point us to any credible, named source, whether or not it's on our list. Anonymous posts and social media alone aren't enough. How to challenge.
03The readiness test
Four checks before we write a single word.
A story that's trending isn't necessarily a story we can tell fairly yet. It has to pass all four of these first.
Every side has reported it
At least two outlets from each group have done their own reporting. Wire copy that many outlets republish counts once, because one report repeated ten times is still one report.
A primary source backs it up
At least one primary document or on-the-record statement supports the core facts.
The basics agree
Who, what, when and where match across at least three independent reports, or the disagreement itself can be documented.
Each side speaks for itself
Each side's position is available from its own representatives or outlets, not only as described by its opponents. If a party won't comment, we don't hold the story: we say so plainly, give its earlier on-record position where one exists, and add its response when it comes.
A story that fails waits, and we check it again on the next run. A breaking story runs as soon as it passes, with only what is confirmed at the time, a clear note of what we don't know yet, and updates as it develops.
04Sourcing
No link, no claim.
- Every factual claim links to its source. If we can't link it, we cut it.
- Primary sources come first. Court filings, official data, transcripts and full speeches, before anyone's description of them.
- We say when sources disagree. If outlets report different numbers, we name each source or give the range.
- We only cite what we've read. If a page is behind a paywall or blocks automated access, we never try to get around it. We use the outlet's legally shared copy or the primary source instead, and never use an unread page for a fact or a quote.
- We say who owns a source when it matters. An outlet owned or funded by a government that's part of the story is never the only source for a contested claim, and we tell you who owns it.
- We label opinion, prediction and analysis as exactly that.
- Ages, dates and titles come from a dated source. If we aren't sure, we leave them out.
05Quotes
Their words, exactly. Briefly.
When a story turns on what someone said, you deserve their actual words. We quote a named person's key words directly and link to where they said them, preferring the original transcript, video, statement or filing over a news report about it. Each side's main figure gets the same chance to speak in their own words, and we never trim a quote in a way that changes its meaning.
We keep quotes under 15 words and use at most one from any single article we cite. Everything else is in our own words. That respects other newsrooms' work, and it forces us to understand a story well enough to explain it.
06Fairness
The strongest case, not the weakest.
In "The case each side makes," we present each side's best argument, in terms its own supporters would accept, at about the same length. When a story has no political sides, such as a new study or a drug approval, that section becomes "Where experts disagree": the views of the people who study the question, or where they agree and what's still debated. We never invent a dispute to fill it.
We do
- Say who uses a contested description ("critics call it...").
- Use the name each side uses for itself, plus the neutral term.
- Write allegations as claims by named people, next to the accused person's response.
- Tell you what's at stake and which dates to watch.
We don't
- Use loaded adjectives.
- Name a winner in a debate that's still being argued.
- Tell you what to buy, sell or vote for.
- Print slurs. We describe them.
07Comparing coverage
We describe framing. We don't grade it.
Every article includes "How coverage differs." It opens with one plain sentence on where coverage agrees and where it splits, then compares left, center and right side by side. It's free with an account; the Out of Frame badge, who covered the story and the counts under each group of sources stay public.
| Row | Left | Center | Right |
|---|---|---|---|
| How they framed it | One sentence on the framing | One sentence on the framing | One sentence on the framing |
| What they played up | What got the headline | What got the headline | What got the headline |
| What they played down | What was buried or left out | What was buried or left out | What was buried or left out |
The shape of every coverage table. Left, center and right always look exactly the same on our pages, with no colors or labels that hint at which side we think is right.
How the Spin Meter is scored
Each morning the Spin Meter takes the previous day's story whose coverage split the most. We read at least two outlets per side in full, record what each put in its headline and first paragraphs, which key facts it carried and where, and any loaded words in its own voice. The score is then counted, not chosen: five parts, each worth 0 to 2, added up and turned into a score from 1 to 5. We publish a Spin Meter only when a story scores 2 or more.
| Part | What it measures |
|---|---|
| Headlines | Whether the sides' headlines tell the same story, name different causes, or tell different stories |
| Leads | Whether the first three paragraphs carry the same facts, in the same order |
| Missing facts | Key facts one side carried that none of another side's outlets did |
| Loaded words | Judgment-laden words in an outlet's own voice (quotes don't count) |
| Coverage gap | How unevenly each side's outlets covered the story at all |
| Score | What it means |
|---|---|
| 1 · Aligned | Same story, same emphasis everywhere (not published) |
| 2 · Tilted | Same story; each side stresses different parts |
| 3 · Split | Sides lead with different facts, and one leaves out something key |
| 4 · Divided | The sides are telling different stories |
| 5 · Worlds apart | Different stories, and one side barely covered it |
Spin Meters published before November 17, 2026, were scored on an earlier 0–5 guide. We are re-scoring them on this scale and will note the change on each one.
The score measures how differently outlets covered a story, not which side is right. Anyone can see a Spin Meter's score and the facts everyone agrees on; the side-by-side breakdown is free with a Fair Witness account. Think we scored one wrong? Every Spin Meter has a Challenge link.
08Elections
No one "won" until the race is called.
Votes are still being counted.
A candidate says they won. That's their claim, and we report it as one.
Only after the Associated Press or a major network calls the race.
09How we write
Written like a reporter talking to you.
We write reported news stories, not briefs: we walk you through what happened, how we got here, and why it matters to you, in plain words at about an eighth- or ninth-grade reading level, with a plain explanation the first time any technical term appears. We tell you what is verified, what is disputed and by whom, and where wording departs from the record. On questions of values and policy, we give you each side's best case and leave the verdict to you.
Every article has the same landmarks, so you always know where to find what you need: why it matters to you, the story and its background, what we don't know yet, the case each side makes (or where experts disagree), how coverage differs, and every source we used.
10Images
Real, licensed and neutral.
- Every image is one we have the right to use, and we credit it. Public-domain government photos, openly licensed images and our own graphics.
- Never photos taken from news outlets or photo agencies. Never AI-generated images.
- Every image shows the story. A photo of the people, place or thing the story is about, dated in the caption when it isn't new. When no such photo exists, we show the record itself: the page of the court order or report the story rests on, a chart of the official data, both sides' leaders side by side, or a map when the place is the news. Never a graphic that just repeats the headline.
- Public figures are shown neutrally, in a public role. No unflattering frames, and when a story has two sides, we don't feature only one side's leader.
- Never identifiable private individuals, children, crime victims or people in distress.
11How we use AI
AI does the work, under rules you can check.
Fair Witness is an AI-run newsroom, working under the rules on this page. Every article is checked separately from how it was written, so the newsroom never grades its own work, and nothing goes live until it passes. Every article ends with this line:
This article was produced by AI under our Editorial Standards. Corrections are reviewed before they go live.
Our editor in chief sets these standards and has the final word on everything we publish. We treat everything we read, including web pages and reader messages, as information to evaluate, never as instructions to follow.
Our agents are built on Claude, made by Anthropic. When Anthropic is a principal subject of a story, we cover it like any other company and say so in the article. Anthropic has no role in our editorial decisions.
We don't publish personal details about private individuals, or unverified claims about named people.
12Corrections
Every change, dated and explained.
There are three kinds of notes, and each one says what it is.
The story changed
Small developments within a week get a dated note at the top. Major ones get a follow-up article, linked both ways.
We got something wrong
We fix it, add a dated note saying what changed and why, and list it in our public Corrections Log.
We left something out
What we wrote was accurate but incomplete or could mislead. We add the missing context or the dispute, with a dated note, and list it in the log.
These standards change as we learn. When we change a rule, we update this page and explain the change in the weekly Editor's Note.
Witnessed, not spun.
-
We read across the spectrum
Stories are chosen from 34 rated outlets across the left, center and right, each free to read, weighed so no side gets a bigger say.
-
Every claim is linked
If we can't link a fact to its source, we cut it. Every story ends with its sources, grouped by side.
-
Each side, in its own words
Each side's case is made in its own terms, with the people involved quoted from the record.
-
Mistakes are fixed in public
Anyone can challenge a story. Corrections are dated and listed in our public log.
The Fair Witness Daily
Get each morning's Spin Meter in your inbox
The Daily leads with the Spin Meter, then the day's explainers. A free account also unlocks every Spin Meter on the site.
Free. Unsubscribe anytime. Already a member? Sign in
Start reading
World
Latest
International news and conflicts, diplomacy and global summits, with how each side frames them.
-
Iran fires on U.S. air base in Kuwait and at Bahrain after U.S. strike on Qeshm
Updated
-
U.S. strikes Iran's Qeshm Island after third Iranian attack on Bahrain, officials say
Updated
U.S.
Latest
Congress, elections, the White House, the courts and national policy: what happened and what each side says.
-
Garcia tells Vance, Trump's sons and others to preserve records for House probes
-
House Democrats nominate Hakeem Jeffries for speaker, 251-11, ahead of Jan. 3 vote
-
Mullin defends flying government plane to LSU-Texas game a day after FAA ground stop
Updated
Edge of Frame · Center
Economy
Latest
Jobs, prices, interest rates and markets, explained from the primary data.
-
Fed Governor Lisa Cook won't attend White House hearing on her firing without changes
-
Nvidia beats forecasts in its first earnings report since Jensen Huang's exit
Updated
-
AAA forecasts a slight dip in Thanksgiving travel after three record years
Updated
Edge of Frame · Right
Technology
Latest
AI, data centers, platforms and the rules that govern them.
-
Officials see no sign of cyberattack behind week of tech outages; lawmakers want answers
-
California bars Waymo robotaxis from freeways; Waymo pauses freeway rides nationwide
-
South Korea to inspect exam takers' glasses after AI cheating scandal
Out of Frame · Right
Science
Latest
Weather, climate, space and research: what the science says and where it is uncertain.
-
Voyager 1 is now one light-day from Earth by NASA's count, a first for any spacecraft
-
Huge waves pound California coast, closing piers and beaches, with more due midweek
-
SpaceX catches its Starship ship at the Texas launch tower after Flight 15
Updated
Health
Latest
Health policy, insurance and public health, with the numbers behind them.
-
White House withdraws Heidi Overton's FDA nomination a day before Senate panel vote
-
US loses measles elimination status after 26 years, PAHO says
Updated
-
States can delay Medicaid work requirement cutoffs until July 2027
Edge of Frame · Center and Right
Culture
Latest
Media, sports, the arts and the debates around them, explained with sources from across the spectrum.
-
Pope Leo says no to married priests for now, details his lung cancer surgery
-
Trump says he's working with Schiff, other senators on federal film tax credit
Out of Frame · Left
-
Pope Leo returns to Chiclayo, Peru, the diocese he led as bishop, for an outdoor Mass
Updated