Skip to content

Editorial standards

How we publish — our editorial standards

Horizon Scanner publishes news analyses and methodology content on EU regulation. This page documents how we select topics, research, review and correct — so readers, supervisors and search engines can assess the depth of responsibility we operate with.

As of:

01Topic selection

We select topics on three criteria: (a) regulatory materiality — publication in the Official Journal, ESA guideline, Q&A update or national supervisory letter with effect on financial institutions; (b) timeliness — news topics are handled within a few days of the primary source publication; (c) reader value — we prefer topics on which our target audience (compliance, risk, IT-security and actuarial functions at European financial institutions) must take operational decisions.

We do not publish clickbait, sensationalised headlines, or speculative predictions about lobbying outcomes. A headline like "X is now banned" is used only when a primary text unambiguously documents the ban.

02Research and source hierarchy

Every article anchors on a primary source: the Official Journal text, the original ESA guideline, the supervisory letter in published form. Secondary sources (law-firm commentary, press releases, analyst reports) are cited but not used as a factual basis — they contextualise; they do not replace the primary text.

Citation rule: Every material claim (a date, a number, an article number, a legal conclusion) carries an inline footnote `[N]` that links to the source list at the end of the article. Source links point, wherever possible, to EUR-Lex via the durable ELI URL (European Legislation Identifier) or the official supervisor domain.

03Four-eyes review before publication

Each article goes through four-eyes review before publication: one author writes; a second reviews sources, dates, legal conclusions and readability. The review is asynchronous in an internal review tool; it is complete before the publish button is pressed. On particularly complex topics (e.g. supervisory convergence or new delegated acts), we seek external expert input — typically practitioners in the industry.

04AI use and human responsibility

We use AI tools to accelerate research, translation and first drafts — but every published article has a human accountable for it: every claim is reconciled by an editor against the primary source, every legal conclusion is double-checked by a second editor, and publication is approved by an accountable lead. AI does not replace editorial judgement; it accelerates the work around it.

This practice is consistent with Art. 26 of the EU AI Act for deployers of high-risk AI systems — although our application in the news-writing process is not high-risk, we hold ourselves to the same standard.

05What we do not publish

We do not publish: personalised investment recommendations (those are MiFID II material); legal advice on individual cases (that is a regulated legal service); customer identifications without express release (confidentiality); disparaging comparisons or direct name-based comparisons to identified competitors (we compare formats, not brands).

06Editorial / sales separation

Our blog articles have no sales objective. They often link at the end to the product — that is a convenience for interested readers, not the editorial purpose. Content is never changed because a customer or prospect requests it. The "book a demo" CTA at the article end is product marketing; the article body is not.