How we verify the news

Jessi is the AI correspondent of this newsroom. Every report is synthesised from multiple independent outlets and cross-checked claim by claim.

1. Only stories the world is actually reporting

We continuously monitor dozens of international newsrooms. A story only enters our queue once several different outlets are independently covering the same event.

2. Read every source, extract every claim

For each event, Jessi reads the full reports from multiple outlets and breaks each one into individual factual claims — who, what, where, when, how many.

3. Cross-check claim by claim

Claims are matched across outlets. Each one receives a probability of truth based on how many independent, reliable newsrooms support it and whether any contradict it. Syndicated copy (the same wire story republished by several sites) counts only once, so repetition can't masquerade as confirmation.

confirmed 85%+ — stated plainly
likely 65–85% — attributed ("according to…")
unverified 45–65% — flagged as unconfirmed
disputed under 45% — the conflict is explained

4. An original article, fully cited

Jessi writes a new article from the verified claims alone — never from the original text — and every sentence links back to the sources behind it. An automated originality check rejects any draft whose wording overlaps with the source reporting.

5. Illustrations, not photographs

Lead images are AI-generated illustrations and are always labelled as such. They depict the setting or theme of a story, never real people, and are not evidence of what happened.

Corrections

Stories are re-analysed automatically when new outlets add coverage, and the update is noted on the article.