AI policy
Where AI is used at hr.cafe, where it never is, and how you can tell the difference on any given page.
Last updated
The short version
A person lived every story here, and a person approved every word of it. AI is used for editorial support — the same work a copy editor does — and never to invent experience.
What is never AI-generated
- People. No contributor is invented. No portrait is generated; profiles use typographic monograms precisely so that no synthetic face ever appears.
- Experience. No story, anecdote, quote or outcome is written by a model. If it did not happen to the person in the byline, it is not published.
- Numbers. No result, metric or reach figure is produced or estimated by a model.
- Responses. Peer responses are written by peers.
Where AI is used
- Shaping five answers into a draft. If you use the five-answers path, an editor may use AI to arrange what you wrote into the format’s structure. It adds no facts. You then approve the exact text.
- Copy editing. Grammar, length, heading suggestions.
- Interview transcription. For the interview path, when you have agreed to it.
- Site building. The code and generative cover art are produced with AI assistance. The art is deterministic geometry drawn from a story’s format — never an image of a person or a place.
How you can check
Every story carries an authorship label in its byline, and the same label appears in the page’s structured data, so it cannot say one thing to you and another to a search engine. The labels are defined on the editorial standards page.
Why we are this specific
A blanket reassurance is worth nothing — every publication says it uses AI “responsibly”. The useful statement is the narrow one: which tasks, and which never. If you find a page where this policy does not hold, tell us and we will correct it publicly.