Editorial Policy
This policy describes how News French Fries researches, writes, edits and corrects what it publishes — and who is answerable for it.
Who is responsible
Leandro Bruzaferro is the editor of News French Fries and is responsible for everything published here. Questions, complaints and corrections reach him at newsfrenchfries@gmail.com. His professional profile: linkedin.com/in/leandro-bruzaferro.
How our articles are produced
Articles are researched and drafted by our newsroom with the assistance of AI tools. Every article is reviewed, fact-checked and approved by a human editor before publication, and the editor who approved it is named on the byline. No article is published without that review.
We disclose this openly. Readers are entitled to know how what they are reading was made.
Sourcing
We prioritise official statements, primary documents, published data and peer-reviewed research. Where a story turns on a figure, a date or a quote, we link to the document it came from so a reader can verify it independently.
Where a claim cannot be traced to a source we can link to, we say so rather than implying a source that does not exist. We do not cite a document we have not read, and we do not attribute a statement to an organisation that did not make it.
Accuracy and corrections
When we get something wrong, we fix it in place and append a note explaining what changed and when. We do not make quiet edits. If you spot an error, email newsfrenchfries@gmail.com and we will look at it.
When an article falls below the standard on this page, we take it down and rewrite it rather than leave it up.
Independence
Our editorial decisions are made independently of advertisers, sponsors and commercial partners. Sponsored content, when present, is labelled as such. No advertiser sees an article before publication.
Opinion and reviews
Opinion is labelled as opinion. Reviews state how the work was accessed.
Conflicts of interest
Personal or financial relationships that could reasonably influence coverage are disclosed. Where a conflict cannot be avoided, the story is reassigned or dropped.
Feedback
We welcome reader feedback and tips. To send a story idea, visit our Send a Tip page.
How we handle numbers
A figure in an article is only as good as the body that published it. Where a story turns on a number — a rate, a cap, a threshold, a count — we link to the primary document, and we check that the version we are citing is the one currently in force. Regulators revise published figures; a number that was correct last year is not automatically correct now.
What AI does here, and what it does not
AI tools assist with research and drafting. They do not decide what is true and they do not publish. A human editor reads every article, checks the claims against their sources, and approves or rejects it. Where AI drafting produces a claim that cannot be traced to a real document, the claim is removed — and if the article cannot survive that removal, the article is removed.
Sources we will not use
- Documents we have not read. If we cannot open a source, we do not cite it.
- The homepage of an organisation standing in for a specific claim. A source has to be the page that actually says the thing.
- Content farms, and articles whose own sourcing we cannot follow.
Taking articles down
Most errors are fixed in place. But when an article’s central claim turns out to be wrong — a subject misidentified, a source that does not exist, an event described incorrectly — patching a line is not honest. In those cases we unpublish the article and rewrite it from verified sources before it returns. We have done this, and we will do it again when it is warranted.
Why we use AI at all
News French Fries is a small publication. Using AI assistance for research and for first drafts lets an operation this size cover more ground than it otherwise could, and it moves the scarce resource — editorial time — to where it actually changes the reader’s experience: checking whether a claim is true, whether the source says what the article says it says, and whether the piece was worth writing at all.
We are explicit about the trade-off. AI is fast and it is confident, and confidence is not accuracy. A language model will produce a fluent citation to a document that does not exist, and it will state a superseded figure with the same certainty as a current one. Those two failure modes are the reason a person checks every article against its sources before it runs, and they are why the source list sits at the foot of each article in public: so a reader never has to take our word for it.
What we do not claim
We do not contact people or organisations for comment, so we do not write that we did. Where an article notes that a company declined to comment or did not respond, that approach was made by the named outlet we credit — not by us. We do not describe ourselves as having obtained documents, attended events, interviewed sources or conducted original data analysis unless that is literally true, and for a publication of this size it usually is not.
What we do instead is read the primary record — filings, rulings, official statistics, governing-body pages and the reporting of outlets that did the fieldwork — explain what it says, and link to it so the reader can check us. That is a narrower job than reporting, and we would rather do the narrower job honestly than imply the wider one.
Advertising and labelling
Part of our revenue comes from advertising displayed on the site. Advertisers have no input into what we cover, and no advertiser sees an article before publication. Ads are served by third-party networks, which means we have only partial control over the specific offers shown; we are not responsible for the products or services they promote. Any content paid for by a third party is labelled as sponsored.