An AI-drafted, human-gated publication

Quill publishes research syntheses and vendor coverage for practitioners working in cyber security and defence. Every article is drafted by a language model against a fixed set of stored sources, verified mechanically, and put in front of a human editor who decides whether it goes out.

Quill is its own publication within the River group. The group’s peer-reviewed journals and its magazine are among the sources Quill reads; they are read-only, and nothing Quill publishes appears on them or is approved by them. The editorial line is Quill’s own.

What we publish

Research syntheses read the peer-reviewed record on a question and report what it actually shows, including where it disagrees with itself. Vendor coverage presents a supplier’s position on the supplier’s own terms — the claims are attributed, never adopted, and the sources behind them are marked as vendor material wherever they appear.

Each article carries the byline of the editor who commissioned and cleared it, the reading time, the number of sources it was grounded on, and how many of its claims are anchored to a specific passage in one of them. Those numbers are the article’s own; nothing about them is rounded in the publication’s favour.

Corrections

A correction is a new revision of the same article at the same address. It passes the full verification gate again before it replaces what was there, and the page then carries a “Revised” line with the date. Quill’s database is the only copy of a published article, so there is no second version anywhere to fall out of step with this one.

Reading Quill

The feed carries complete articles rather than teasers — the audience is practitioners and there is no advertising to protect. The verification page sets out exactly what is checked before anything is published, and what those checks do not cover.