Thursday, November 19, 2026Thu, Nov 19 · Briefs edition · 26 stories

Editorial standards

Trust is the product. These are the rules every DexScience story has to pass before it is published.

1. Primary sources only

We report from the organisations that did or published the work: space and science agencies, observatories, peer-reviewed journals, universities, research institutes and national laboratories. We do not rewrite other news outlets' journalism, opinion pieces or commentary. Company research blogs are used for announcements by that company and are labelled as such.

2. Facts, not copies

Science is open, but writing is not free to copy. Our articles are written in our own words from the facts in the sources; we do not reproduce press releases or papers. An automated originality check rejects any draft that shares long passages with its source. We publish no direct quotations, so no one can be misquoted. Every article links to every source it used.

3. Every claim checked

Before publication a fact-checking step compares each sentence with the source material: numbers, units, names, dates, institutions and the strength of each claim. Problems are corrected and checked again; if a story still fails, it is not published.

4. Honest certainty

Hedged findings stay hedged. Preprints are clearly labelled as not yet peer-reviewed. Results in animals or cells are never presented as results in people. Health stories report research; they are not medical advice.

5. Images

Images are taken from the original source and stored on our own servers only when the licence allows reuse, such as public-domain U.S. government works or Creative Commons Attribution licences. Each image carries its credit and licence.

When a source's images cannot be reused, we look for a real related photograph or scientific image of the same subject in the NASA Image and Video Library or on Wikimedia Commons. We accept only public-domain, CC0, CC BY and CC BY-SA files, credit the author, never use AI-generated pictures, and label the image as related rather than from the study. Only when no suitable image exists do we use a clearly labelled DexScience illustration.

6. Choosing stories

On major days (Monday, Wednesday, Friday, Sunday) each section leads with its most significant stories; on the other days we publish smaller but relevant results. Significance is judged on the finding itself, not on how it is promoted. Corrections, events, awards, job posts and marketing are excluded.

7. Use of AI

DexScience uses AI models (currently Anthropic's Claude) to filter candidate stories, draft articles, fact-check drafts against their sources and describe images for accessibility. Each step works only from the linked source material. The process is supervised by the site's editor, who reviews reported errors.

8. Corrections

If we get something wrong we fix it, note the correction on the article, and keep the record. Report errors through the contact page.

9. Advertising and independence

Advertising, when present, is clearly marked and never influences which stories we cover or what they say.