How we summarise news

Where stories come from

We follow a short list of sources: AI labs and research teams, established technology publications, and organisations that track AI policy. New articles from these sources arrive through their public news feeds. We choose the stories that matter most for people who use, build or are affected by AI, across five topics: Models, Research, Tools, Policy and Business.

How a summary is written

1. Read the original. Each summary starts from the full original article. If we can’t read it, we don’t write the story.
2. Draft in our own words. The summary is drafted with help from AI tools, in plain language and in our own words. We don’t copy sentences from the source, and we only quote a few words when the exact wording matters.
3. Check every fact. Every number, name, date and claim is checked against the original article. If we can’t find it there, it comes out. We never estimate or add numbers from elsewhere.
4. Publish by hand. A person reviews each story and decides whether and when it goes live. Nothing is published automatically.

What every story contains

• A headline and a one-sentence summary that say what happened.
• At a glance: up to three key numbers from the source, what’s new and who it affects.
• The details: the context you need, without repeating the numbers.
• Why it matters: our own short analysis of what the news could change next, and any open questions or limits. This part is our view, not the source’s.
• The original source: the publication, the original headline and a “Read the original” link. The full reporting always lives there.

Corrections

If a summary gets something wrong, please tell us through the contact form, with the story and what needs fixing. We’ll check it against the original source and correct the story.