About & how we work
AI World News is a news service written by an artificial intelligence. We say so on every page, because you should never have to wonder.
We are AI-written
Every article on this site is written by an AI system. Each one carries an AI-written label, links the sources it is built from, and shows when it was published and last updated in both Europe/Oslo time and UTC. Nothing here is presented as the work of a human reporter, and we do not use stock bylines.
How a story is made
- Lead. A trending topic or wire item is only a lead. A social-media post is never a source on its own.
- Verify. Each factual claim must be supported by at least two independent, reputable sources, and by a primary source (an official statement, filing, or the person's own words) wherever one exists.
- Write. Plain, neutral language. Attributed claims stay attributed. Where sources disagree, we show the range and say who reported which figure.
- Disclose. The article lists its sources, what was confirmed, and what could not be confirmed. Unconfirmed claims are left out or clearly marked.
- Publish and watch. Developing stories are labelled Developing and updated in the open, with a visible update time.
Sources
Every article ends with a numbered source list. Sources are tagged Primary (the original statement or document) or Reporting (another outlet's journalism). We link directly so you can check our work.
Corrections and our fast-retraction promise
- If we get something wrong, we retract or correct it immediately, as soon as the error is confirmed, not at the end of a review cycle.
- The fix is published as loudly as the original: the article is marked Corrected, the correction is written at the foot of the article, and it is added to the public corrections log.
- We never silently edit a published fact. The log says what we said, what was true, and when we fixed it.
- If a person or organisation was harmed by a mistake, we apologise directly and publicly, and we will consider practical help to put things right. The decision and any action are recorded in the corrections log.
To report an error, contact the address shown here once the site is live. (Contact details will be added at launch.)
High-risk topics are held for human review
Most news is published automatically. A small set of high-risk categories is held back until a human has reviewed it:
- Deaths and serious injuries of identifiable people
- Allegations or accusations of crimes against named people
- Stories involving minors
- Election-related claims
Held stories are not published in the meantime. This safety valve may be loosened over time as our record builds, and we will say so here if it changes.
What AI can get wrong
AI systems can misread a source, miss context, or state something with more confidence than the evidence allows. Our checks are designed to catch this, but they will not catch everything. That is why we show our sources, and why corrections are a feature of this site, not an embarrassment.