Editorial and AI policy

This page describes exactly how the site produces what you read. It is deliberately specific, because the process is automated and you deserve to know what that means.

How stories are selected

Public feeds provide candidates, not publication-ready facts. New candidates must have a known recent source date and describe a concrete change relevant to independent work. Publisher diversity and daily volume limits prevent one source from dominating the wire. The site may publish nothing when no candidate passes.

Full-source reading and review

The system retrieves the source page, respects its crawler restrictions, and extracts the article text. Inaccessible, overly long or insufficient sources are deferred or rejected; feed excerpts alone cannot pass the current publication workflow. Where available, one linked reference from a recognised official or research domain is also retrieved. A link is not automatically treated as corroboration.

A model drafts the report, practical implications and limitations using only retrieved evidence. Key claims must include internal excerpts that match the source text. A separate model call checks the complete draft against that evidence. Unsupported claims, invalid output and excessive source-wording overlap prevent publication. Technical failures retry with backoff; repeated failures enter a cooldown.

This is an automated source-consistency check, not independent fact-checking. A source can itself be mistaken, and automated checks can miss errors. Community claims are attributed; interpretation is labelled. We do not claim hands-on product testing.

Dates and corrections

We distinguish this site's publication and update dates, the source publication date, and an event date only when the evidence establishes it. Unknown dates are shown as unknown. Earlier feed-based summaries are reviewed in a separate automated queue. Successful revisions preserve their URL and a private prior-version record, with an on-page update note. A failed attempt to prepare a revised article does not remove the existing URL from search discovery. The previous edition remains accessible with its review status, canonical URL, original publication date and sitemap entry; new candidates are withheld until their checks pass.

Machine-readable editions

Every article automatically exposes Markdown and JSON alternatives with the same content, dates, review status and evidence links. A paginated directory at llms-index.json covers the full archive. Feeds and sitemaps are generated from the publication database. Search notifications use a durable queue with retries; acceptance and indexing are not guaranteed.

AI disclosure

Every item on this site is drafted with AI assistance and published automatically without a human reading it first. Each article says so, on the page, next to the source link. The publication — not an invented journalist — is credited as the author in our metadata, because no person writes these summaries.

Sources and copyright

We summarise; we do not republish. Facts are not copyrightable, expression is, and our rules exist to keep us on the right side of that line. Each page names the original outlet and links to it. We do not copy images. If you are a publisher and believe an item crosses a line, write to hello@freelancenews.online and it will be amended or removed — typically the same day, and without argument.

Corrections

Automated publishing means mistakes are possible: a misread headline, a summary that is technically accurate but misleading, a story that turns out to be false at the source. Write to hello@freelancenews.online with the URL and what is wrong. Corrected items are updated in place with a note. An unsuccessful automated source review is disclosed on the page; it does not silently erase the previous report.

Independence and disclosure

The site's publisher also runs SnipeWork and a design practice, and both are promoted here in clearly labelled blocks. Commissioned editorial comparisons may also cover these products, with the ownership relationship disclosed in the article. Nothing on this site is paid placement, and coverage is never traded for anything. Where a story touches a product the publisher owns, that relationship is stated in the item itself.