Skip to content
All categories
Worldwide

AI disclosure

We use AI models in a limited way, and we want to be clear about where.

What is data, and what is not

  • Source data: topic names, traffic buckets, start times and news headlines come from Google Trends and Wikipedia and are shown as published.
  • Our calculations: ranks, rank changes, graphs, "trending for" times and counts are computed by our own code from the data we collect. No AI model produces these numbers.
  • AI-assisted decisions: the category of a topic, and a sensitivity check (whether a topic is about a death, a crime, a private individual and so on).
  • AI-assisted text: draft explanations of why a topic is trending (approved by a person before they appear) and the content ideas you can request.

How AI is used

When rules and Wikipedia's categories cannot decide a topic's category, or to check whether a topic is sensitive, a language model is given the topic's name and a few related headlines and returns a category and a yes/no answer. We validate every answer against a fixed list and keep a confidence value; low-confidence results are reviewed by a person. Topics flagged as sensitive are listed by name only.

"Why it is trending" texts

For some topics an AI model writes a one- or two-sentence draft from the news headlines we hold. A draft is never shown to visitors. A person reads it, edits it if needed and approves it; only then does it appear on the topic page, together with the reviewer's name, the date and a link to a source. Topics that are sensitive or not yet checked never get one.

Content ideas

The "Content ideas" button asks an AI model for five suggestions (a video hook, a post title, a blog headline, a question to answer and an angle to avoid) based on the topic name and a few headlines. They are generated when you press the button, are not reviewed by a person, and may be wrong or unsuitable, so check them before you use them. They are not stored in the page and are cached for the day.

The models are provided by Google (Gemini) and, where configured, Groq. We do not use AI to write articles, images or the rankings themselves.

Mistakes

Automated classification is imperfect. If a topic is under the wrong category or is wrongly flagged, tell us through the contact page.