01 — Sourcing
We link primary sources: release notes, documentation, filings, papers and the product itself. When we rely on a person, we say who they are and how they know.
We grant anonymity only when a source would face real harm, and we say why in the story. Anonymous claims are corroborated before publication.
02 — Original testing
We pay for the subscriptions we test. Vendors never see results, scores or copy before publication. Hands-on pieces record the exact versions and hardware under “Tested on”.
Benchmarks follow a public, versioned methodology, and raw data is published under CC BY 4.0. Read the methodology.
03 — Updates
When a story changes materially — new facts, new figures, a changed conclusion — we add a timestamped note at the top and update the “Updated” time. Typo fixes and formatting changes don’t get a note and don’t change the date.
04 — Corrections
Errors of fact are corrected as soon as we confirm them, with a note at the end of the story saying what was wrong. Significant corrections are also listed on our corrections page.
Write to corrections@9to5ai.tech; we usually fix errors the same day.
05 — Headlines
Headlines say what happened. Search headlines lead with the subject; social headlines are specific and never withhold the point to earn a click.
06 — Use of AI
Every story is reported and written by a person. We may use AI tools to transcribe interviews, check translations and run the tests we describe. When a tool materially shaped a story, the story says so.
We never publish AI-generated articles, and we don’t use AI to generate images presented as real.
07 — Independence & conflicts
No company can pay for coverage, placement, a score or a ranking. Staff disclose investments and relationships on their byline pages, and don’t cover companies they hold stakes in.
08 — Embargoes
We honour embargoes we agree to in advance. We don’t trade favourable coverage for access, and we don’t agree to show copy to the companies we cover.