Editorial policy

How patents are selected

Candidate patents come from watchlist queries and CPC-code filters against public patent data. An operator reviews each candidate and decides whether it is worth covering before anything is drafted — nothing is published automatically from a watchlist hit.

How articles are drafted and reviewed

Once a patent is selected, a language model drafts an article from that patent's title, abstract, claims, and figures — the same source document a human would read. The model does not invent citations, dates, inventors, or figures; every fact in the Sources & credits block is pulled directly from the patent record, never from the model's own output.

The draft then goes to a human editor, who reads the full article, checks its description of the patent's problem and claims against the source document, and confirms the attribution and figure credits are correct before approving it. An article that has not been through this review is marked noindex and is not reachable from any public page on this site — search engines, the sitemap, and direct links all exclude it until it is approved.

How figures are sourced and credited

Every figure shown with an article is reproduced from that patent's own published drawings. Some are re-styled for readability on this site; none are AI-generated depictions of the invention. Each figure's caption states exactly which is true — see the article's AI-assistance notice and Sources & credits section for the specific wording used for that figure.

Corrections and takedown requests

If an article misrepresents a patent, misattributes a figure, or you are a rights holder with a concern about how a figure from your patent is used here, contact us at corrections@contentgenerator.example. Include the article URL and a description of the issue. We review every report and correct or remove content promptly where warranted.