UK insurance news

How CoverWire works

CoverWire is created using AI. This page sets out how a story gets from an announcement to the site, and where people are and are not involved.

Each weekday

  1. Selection. An AI system reads the day's insurance announcements and news and picks the stories to cover, following rules set by the founder, David Tucker.
  2. Research. For each story, an AI system collects the sources: the announcement itself, regulatory or court documents and other reporting. It records the facts with links, direct quotes, what is not known and what must not be said.
  3. Writing. An AI system writes the article from that research, in CoverWire's house style.
  4. Checking. A separate AI step copy-edits the article and checks it against the research, then each claim is checked line by line against the sources. Automatic checks block publication if sources, a description or a credited image are missing.
  5. Images. Every story has a credited photograph: a company's press image where one is available, otherwise a photo from Unsplash or a Creative Commons photo from Wikimedia Commons. Where no photo of the subject is available, a relevant general scene is used and the caption says what it shows. CoverWire does not use AI-generated images.
  6. Publishing. Most news stories are published automatically. Stories that make negative claims about a named person, report unconfirmed deals, or rest on weak sourcing are held back.

Where people are involved

David Tucker sets and changes the rules, reviews how the system is performing and approves longer analysis pieces and the newsletter before they are published. He does not read every news story before publication. CoverWire does not claim human review for an article unless it has had one.

What CoverWire does not do

  • It does not interview people. Quotes come from published sources and are attributed.
  • It does not publish paid-for or sponsored content.
  • It does not use AI-generated images.

Mistakes

AI makes mistakes, and so do the sources it relies on. If you spot one, email [email protected]. Corrections are dated and say what changed. See the corrections policy.