AI News
What changed, and how much to trust it.
Every item is summarised in plain language, labelled with how confident we are, and linked to its original source. We never republish — we point you at the primary source.
How we label sources and confidence
Source types
- Official announcement
- The organisation the story is about, speaking for itself.
- Research / paper
- A paper, preprint, or benchmark release, published by its authors.
- News report
- Independent journalism by a publication with an editorial process.
- Expert analysis
- A named individual with relevant standing interpreting something.
- Social media update
- A post on X, LinkedIn, or similar — confirmed only when first-party.
- Community discussion
- Reddit, Hacker News, Discord — aggregate practitioner observation.
- Rumour / unverified
- A circulating claim with no adequate sourcing, labelled as such.
- User submitted
- A reader-supplied link, before or after editorial review.
Confidence levels
- Confirmed
- Multiple credible sources agree, including a primary one, or the primary party stated it directly.
- Likely
- Credible reporting, not yet directly confirmed by the party involved.
- Developing
- Real activity is visible but details are moving or contested.
- Unverified
- A circulating claim with insufficient sourcing, shown so it can be correctly labelled.