How — and how not — we use AI tools in producing MatchaNews.
The first version of this site was built with AI tools, and the AI was allowed to invent placeholder news: fake bylines, a fake research citation, fake interview quotes, and fake ratings attached to real cafés and shops. We removed all of it in August 2026. That happened because we didn't have a policy for what AI could and couldn't do — this page is that policy, written so it can't happen again and so readers know exactly where AI sits in how MatchaNews gets made.
Every article carries a named human byline, and that person is responsible for checking every factual claim and source link before anything goes live — regardless of whether AI was involved in drafting it. An AI-assisted draft is a starting point, not a finished article.
We use AI tools in our production workflow, as described above. We're not labeling individual articles as "AI-assisted" or "human-written," because under this policy every published article is both — AI may help with research and drafting, but a human editor verifies and takes responsibility for the result. See our Editorial Policy for how that verification works in practice.
If something we've published looks off — a stat that seems wrong, a quote that seems too smooth, anything that reads like it wasn't actually checked — tell us. Contact details are in the site footer and on our Editorial Policy page. We'd genuinely rather be told.