Editorial standards
How stories are selected and grouped, where AI is used, and what we do when we get it wrong.
Last updated 13 August 2026
What we are
News Clarion is an aggregator, not a newsroom. We employ no reporters and produce no original journalism. Every factual claim on this site originates with a publisher we link to, and the link is the point — we exist to send you to their reporting, not to replace it.
How sources are chosen
We read the public RSS feeds that publishers provide for syndication. Sources are selected for editorial reliability, and each carries an internal credibility rating used when deciding which headline best represents a story: wire services and papers of record rank above national outlets, which rank above specialist and trade press.
We filter out material that is not journalism even when it arrives through a news feed — newsletter promotions, horoscopes, puzzles, and earnings-call transcripts. The test is what a thing is, never where it came from.
How stories are grouped
Headlines are compared automatically to detect when several publishers are covering the same event, and those articles are grouped into one story. The process is algorithmic and imperfect: occasionally two distinct events are merged, or one event appears twice. Tell us when you spot it.
Within a story, sources are ordered by credibility rather than by recency or by any commercial relationship. We have no commercial relationships with the publishers we link to.
Where AI is used, and its limits
Three things on this site are AI-generated, and each is labelled where it appears: the story synthesis, the analysis of how outlets framed an event, and the background on topic pages.
The model is never given article text. It sees only headlines, publisher names, and the short summaries publishers put in their own feeds — the same material we are permitted to store. It is instructed to draw only on what it is given rather than on its own knowledge, because a model recalling background about a live news topic is exactly where a confident, wrong claim comes from.
Outputs are checked automatically before publication and rejected if they are too short, too long, attribute a view to a publisher not in the source material, or use time-relative phrasing that would silently become false. Rejected output means no summary appears — we would rather show nothing than something wrong.
AI output can still be wrong. Where it conflicts with a linked article, the publisher is right and we are wrong. Treat these summaries as a guide to the coverage, never as a substitute for reading it.
Framing analysis
The “how outlets framed it” section describes editorial choices visible in the headlines of a single story — what an outlet led with, what it emphasised. It is deliberately not a judgement of any publisher's reliability, politics, or motives, and the model is instructed never to characterise an outlet as biased or agenda-driven. Publisher names shown there are always verified against the actual sources of that story.
Corrections
Email app.realm.ar@gmail.com with the page and the problem. We aim to respond within two working days.
Where the error is ours — a bad grouping, a wrong summary, a misattributed framing — we correct or remove it. Where the error is in a publisher's article, we cannot correct their reporting; take it up with them, and tell us if you would like the item removed from our index in the meantime.
Advertising
Any advertising on this site is clearly labelled and kept separate from editorial content. Advertisers have no influence over which stories appear, how they are grouped, or how they are ranked, and we accept no payment for inclusion or placement.
See also our privacy policy and information for publishers.