USA • Tuesday, July 7
Transparency

AI Content Disclosure

Last updated: February 24, 2026

US News Desk uses large-language-model (LLM) assistance during the drafting stage of every editorial. We believe transparency about this is more important than pretending otherwise, so this page describes exactly how AI enters our workflow, what safeguards we have in place, and where the human editors always sit in the loop.

Where AI is used

  • Topic ideation — we use aggregated US Google Trends and our own editorial calendar to pick topics. AI is not used to select what we cover.
  • Research synthesis — we compile a "grounding bundle" of public-source reporting on a given topic and feed it to the model. The model produces a first draft strictly within that grounding bundle. It does not invent facts, quotes, or figures.
  • First draft — the LLM writes a structured first draft using the format we define in our style guide (deck, "the story so far", "why this matters", "editorial analysis", "what to watch next", "for global readers", "the bottom line").

Where AI is NOT used

  • Fact-checking — every named entity, numeric claim, date, and quotation is checked by a human editor against the source list before publication.
  • Editorial judgement — the angle, framing, tone, and any comparative commentary is edited (and often rewritten) by a human editor to reflect our editorial voice, not the model's default.
  • Corrections — corrections are handled by a human editor and applied in place with a visible update note.
  • Comment moderation — reader comments are moderated by humans.

Model & provider

We currently use Google's Gemini 3.1 Pro Preview via a managed API. We may switch models in future; when we do, this page will be updated.

Editorial review stamp

Every editorial ends with an "Editorial review" line showing the human reviewer's name and the review date, plus a list of "Cited sources" that were checked against the piece. If any editorial ever shipped without a review stamp, please report it to saimuralidharmaheedhara@gmail.com and we will investigate and correct it.

Original imagery

Editorial illustrations are licensed from Pexels under the Pexels license. Photographer credits appear on every editorial. We do not use AI-generated imagery on the site.

Reader control

If you would like an editorial removed, corrected, or annotated, please write to us. See our Editorial Policy for the full corrections workflow.

Made with Emergent