For publishers

ovr.news surfaces evidence-based stories from global news sources. We aim to drive readers to your original work, not replace it. Here's exactly how we handle your content.

What we do with your content

• We read your public RSS feed and, when the excerpt is too short, retrieve the full article text from your website to evaluate and summarize it.
• Our AI writes an original summary in new words. We do not republish your text. Scientific and research sources receive a summary format adapted for research findings.
• Article text is processed by open-source models on local hardware.
• We keep the text we collect: in our off-site backup, and in the samples we use to measure the accuracy of our summaries. We don't use it to train AI models.
• Article pages show a credibility band for your publication and say who gave it: an outside rater such as Media Bias/Fact Check or NewsGuard, or our editor.

How we attribute

• Every article prominently shows your publication name and links directly to your original article ("Read on yourdomain.com").
• Article pages set <link rel="canonical"> pointing to your original URL, telling search engines your page is the authoritative version.
• Thumbnail images are loaded directly from your servers. We don't rehost them.

How we collect, and its limits

• We have no licence from news publishers. We rely on public syndication feeds, a named source and a link back to you. That is our position, not a permission.
• We don't log in, and we don't get past paywalls, consent walls or bot challenges. Our collector presents itself as an ordinary web browser. If a feed request fails, it may load your homepage once and try again. For three feeds that refused a plain request, it imitates a browser more closely. That is the one place we work around a refusal.
• We read machine-readable reservations of text and data mining rights (tdmrep.json, TDM headers, noai) under Article 4 of the EU Copyright Directive. A new source that carries one is declined. Nineteen publishers we already carried have since made such a reservation (under 1% of what we collect). We still collect them, as a known risk, and we will ask each of them for consent.
• General robots.txt rules are not checked when we collect.
• No advertising or commercial use of your content. ovr.news is non-commercial. If that ever changes, any revenue model will be transparent, privacy-respecting, and aligned with our editorial values.

Editorial context

Your articles appear in a curated feed that also includes non-article context cards: historical figures, verified progress statistics, and notes about our methodology. These are ovr.news editorial content, visually distinguished from your articles, and always cited to external sources. They are not commentary on your reporting.

Opt out

Want your content removed from ovr.news? Email info@ovr.news with the subject "Publisher opt-out request". We'll remove your sources within 48 hours. No questions, no hassle.

Full transparency

The scoring filters, dimension definitions, and ranking formula that determine how articles are selected and ordered are public.

• How we rank stories — the full ranking rules, including the age bands and the five fresh slots per lens.
• LLM Distillery on GitHub — scoring filter definitions for each lens.

Last updated: April 2026