Articles for your site. Every claim shows its source.

No. The engine reads several outlets covering the same story, extracts the claims each one makes, checks each claim across them, and writes a new text in its own words. It does not copy sentences or paragraphs. Every article lists the sources it drew on, so readers and editors can check the original reporting. A similarity check runs on each article and flags any passage that comes too close to a source.

Disagreements are kept, not averaged. When outlets give different figures or accounts, the article attributes each version to its outlet ("Reuters reports 15 countries; Politico puts it at 14") and the Corroboration tab shows the conflict. A claim counts as corroborated when two or more independent origins support it, or an official source and an outlet do. Any other claim is underlined in the text. You add a source or keep the flag with a note, and a kept flag stays underlined.

You do. Articles are your work product; you may publish, edit, sell and license them without attribution to Brenzuri. The source list belongs with the article and we ask that you keep it when publishing. We do not train models on your briefs or articles, and we do not reuse your articles for other customers.

It is built not to. The writer only receives the text of the grouped sources and is instructed to write from them alone. Each claim is checked across those sources before writing, and a second model reads the finished text back against them. A claim that is not corroborated is still written, underlined, and in the export it carries †. Which AI model wrote each version, and when, is recorded in the Versions tab.