A standing study · updated monthly
Who actually writes the world’s news?
We checked 1,650 news articles from 632 outlets in 105 countries to see who actually reported them. Most were not doing their own reporting.
For 1,485 of the 1,650 we could establish who did the reporting. Every figure below is a share of those 1,485, give or take about four percentage points.
First, what a wire agency is
Reuters, Agence France-Presse and the Associated Press are wholesalers of news. They keep reporters in places most newsrooms cannot afford to staff, and sell what those reporters file to everyone else. A paper that subscribes can print the report under its own masthead, often word for word, with a credit line at the bottom that most readers never look for. That is normal practice, and it is how a paper in Lagos covers a flood in Manila. It also means one reporter’s account of an event can be the account in a hundred papers at once. This study counts how often that happens.
Half the time, two newspapers from different countries were printing the same agency’s words — under two different mastheads.
This is the reason “read more than one source” is not, on its own, useful advice. Two sources can be one source wearing two hats.
Who supplies the reporting
Of the 822 articles carrying someone else’s reporting, these are the suppliers that turned up most often. 62% came from a wire agency; the rest were credited to another named organization — usually a newspaper or broadcaster, sometimes an official source. The number beside each bar is how many of the 822 we traced back to that supplier.
No country’s press is exempt
The obvious next question is whether some countries’ media do more of their own work than others. Barely, and only at the extremes.
How to read this. Each bar is one country. It shows the share of that country’s articles that were someone else’s reporting rather than the outlet’s own work. A bar further to the right means more relaying; further left means more first-hand reporting.
Each bar is a range rather than a single figure. We have only a few dozen articles from each country, and a few dozen cannot pin a number down exactly — the bar covers where the true share very likely falls. The ranges overlap so heavily that ordering the countries would invent a precision the data does not have, so they are listed alphabetically.
Only countries with at least 20 articles we could check appear here; the count beside each bar is how many that country contributed. That even the most independent press here sits near 39% is the point: leaning on the agencies is a feature of how international news is produced, not a national characteristic. A high share is not automatically a failing either — a small outlet publishing Reuters on a distant war is doing something sensible. It matters when every outlet available on a story is doing it at once.
Who was still doing their own reporting
Outlets we could check at least five times, ranked by how much of what they published was their own work.
Three stories you can check for yourself
Everything above is a count. Here is what it looks like on a single story — three pieces of analysis in which one organization supplied the reporting for outlets in three or more different countries at once.
These are not hand-picked. The page takes the most recent stories meeting that bar — two supplied by a wire agency, one by another single source the rest relayed, never the same supplier twice. Every source listed on those pages links to the original article, so you can check each one yourself.
2026-09-25 · 3 outlets, one source: BloombergF-35 parts from Australia reached Chinese hands after Hong Kong diversion🇦🇺 WAtoday Australia🇮🇳 Hindustan Times India🇺🇸 Townhall United StatesRead that analysis →
2026-09-28 · 3 outlets, one source: Ynet News interviewOne interview quote travels from Israeli podcast to Palestinian warning and Arab analysis of West Bank escalation🇶🇦 Al Jazeera Qatar🇹🇷 Anadolu Agency Turkey🇬🇧 Middle East Eye United KingdomRead that analysis →
Why we counted this
Our daily open-source intelligence on world events exists to set national coverage of the same event side by side. That only tells you something if the coverage is genuinely different. So we check, on every story, whether each outlet wrote what it published.
The answer, more than half the time, is no — and when it is no, the reporting usually traces back to one of three organizations. Comparing outlets without checking this produces a false sense of corroboration: the same account, counted several times.
Method, and what this does not show
Every article cited in our analysis is checked for a byline or credit naming a wire agency or another publication. An article counts as someone else’s reporting only when such a credit is found, and as the outlet’s own only when its own reporting is identified. Where neither can be established the article is left out rather than guessed at.
- Window: 2026-08-01 to 2026-10-01. 1,650 articles across 349 pieces of analysis.
- Excluded: 165 articles (10%) where we could not tell either way — paywalls, pages that would not load, and outlets that credit nobody. Every percentage here is out of the 1485 we could check.
- Not a random sample. These are the articles our editorial process selected for comparison, which favors breadth of country over breadth of outlet. It describes the press as it presents itself on contested international stories, not the whole of journalism.
- A credit is not a verdict. Publishing an agency’s reporting is normal practice and often the right call. The finding here is about concentration, not laziness.
This study is rebuilt automatically at the start of each month, as new analysis is published — every figure above moves as the archive grows. Last updated 2026-10-01.
More from our research
Our other standing studies. Each is built from the same archive and rebuilt as it grows.
We do this check on every story we publish.
Several times a day, we take one story moving through the global press and set out how different countries are reporting it — checking, for every article we cite, whose reporting it actually is.