Methodology & AI
Methodology & AI
Evidence first. Each piece is built from identifiable sources, preferably primary and official (government data, filings, official statements). We separate facts, analysis and opinion, state uncertainties and include relevant counter-evidence. Neutrality, for us, means weighing evidence — not giving equal space to unequally supported claims.
How AI is used. VERXERA uses AI assistance for research, drafting and technical checks under written editorial rules. A human editor is accountable for VERXERA's editorial rules and for every publication. Not every piece is reviewed by a human before publication; pieces with heightened legal or personal impact require prior human approval. The public article record must describe only checks actually performed for that version.
Our workflow operates as follows: AI systems research, draft and run technical checks; before release, each piece is reviewed by an independent AI system from a different model provider than the one that drafted it; a human editor reviews a weekly sample of published pieces. A label stating independent AI review appears only when an archived review matches the exact published version. Pieces published before this workflow began carry only the notice of the checks actually performed on them.
New coverage with heightened legal or personal impact is retained under our stricter internal release controls. Publishing these information pages does not enable that coverage. Corrections or withdrawals approved by the responsible editor to mitigate an existing issue follow their own preservation and decision procedure.
Breaking news. Urgent coverage (e.g., hurricanes, major emergencies) requires at least two independent official or primary sources, is labeled "Developing" with update times, and is re-checked frequently during the first 24 hours.
Accountability. Editor responsible: JJ Vazz. Errors are corrected under our Corrections Policy.
What we don't do. We don't present AI systems as human journalists, don't invent sources or quotes, and don't publish AI-generated images as if they were photographs of real events.