Urgent.News

650+ sources. One page. See who else covered it.

Editions

Standards

Editorial Standards

What this site does, what it does not do, and how to tell when it is wrong.

Last updated 9 August 2026.

What this is

urgent.news reads the public feeds of 650+ news sources, groups the many reports of one event into a single story, and writes its own short account of each. It employs no reporters and breaks no news. Every claim on this site originates with a publisher we link to, and the value we add is corroboration — showing how many outlets carried a story and who filed first.

Everything here is machine-written

Summaries, headlines in the retold view, translations and section labels are produced by language models without a human reading them first. That is disclosed on every story and in an ai-generated meta tag on every page. One operator runs this site; there is no copy desk, and pretending otherwise would be the first inaccuracy.

The practical consequence: a summary here can be wrong in ways the original is not. Models drop qualifiers, harden hedged claims, and occasionally attribute a statement to the wrong party. Where a story matters to you, follow the link and read the publisher.

How stories are selected

  • Ingestion is automatic. Sources are chosen by the operator; individual stories are not. Nothing is hand-picked onto the front page.
  • Prominence follows corroboration and recency — how many outlets filed, and how recently — not clicks, dwell time, or anything about you. There is no engagement optimisation here because there is no measurement of engagement to optimise.
  • Some sources are excluded from the lead. Aggregators, commentary blogs and single-topic feeds can appear in a section without being eligible for the front-page lead, so the top of the paper stays reported news.

Where it goes wrong

Stated plainly, because a standards page that lists only virtues is marketing:

  • Section misfiling. Stories are assigned to a desk by reading the article body. Before 15 August 2026 most inherited their publisher's default section, which put some stories under headings they had no business under. The archive has since been reclassified, but new stories can still land in the wrong section.
  • Clustering errors. Two different events with similar wording can merge; one event covered in two languages can fail to merge and appear twice.
  • Translation drift. Foreign-language stories are summarised into English rather than translated. The summary is ours; the reporting is the publisher's.
  • Coverage counts are a floor, not a census. "23 outlets" means twenty-three of our sources filed. Outlets we do not read are not counted.
  • Stale feeds fail quietly. A publisher whose feed breaks simply stops appearing, which can make a story look less covered than it was.

Corrections

When something is wrong we mark it rather than quietly editing it. Published corrections are listed at /corrections, newest first, and a corrected story links to its correction. If the corrections page is empty, that means no correction has been published — not that none was needed.

Found something wrong? Report it. You do not need to leave an address.

Independence

No advertising, no sponsored placement, no affiliate links, no accounts, no payment from any publisher for inclusion or prominence. Nobody has ever paid to appear here, and there is no mechanism by which they could.

Publishers

We read only public feeds, write our own summaries rather than reproducing article text, always link and attribute, and honour robots.txt and machine-readable opt-outs. To have a feed excluded or an item removed, use /dmca or email legal@urgent.news.