Trust Center · Methodology v1.2

We're building the most trusted "see every perspective" platform in the world.

Trust isn't a marketing line — it's a checklist. This page is the checklist. Everything Prism does to earn your trust is documented below, in plain English, with no soft language.

11+
Outlets indexed
9
Stories today
108
AI insights logged
0
Words written by Prism's editors
Our principles

Four commitments. No exceptions.

No editorial bias

Prism does not write its own news. We aggregate and compare what real outlets publish. The only opinion we hold is that every reader deserves to see all of them.

Every source labeled

Every article carries its outlet name and our Left / Center / Right bias label drawn from independent rating bodies (AllSides, Media Bias/Fact Check). Disagree? See our methodology and corrections page.

AI is disclosed, never disguised

AI-assisted features (Common Ground, Flip the Script, Story Summaries) wear an "AI-assisted" badge in plain sight. We never invent quotes, fabricate sources, or attribute words to real publications.

Corrections, not deletions

Mistakes are public. If we mislabel a source or an AI output misfires, we correct in place with a dated note — never silently edit and pretend it never happened.

AI disclosure

Exactly where AI is used — and where it isn't.

Prism uses one AI model: openai/gpt-4o-mini, accessed via the Emergent platform. Here is every place that model touches the product.

Common Ground
GPT-4o-mini extracts factual claims that appear across multiple ideologically-distinct sources, then surfaces what each side adds on top. Never invents facts.
Flip the Script
GPT-4o-mini rewrites the same factual scaffolding in three editorial voices (Left, Center, Right). Headlines are clearly labeled as AI rewrites — never as real outlet headlines.
Perspective Summary
GPT-4o-mini summarizes what each ideological bucket's coverage emphasized. We do not paraphrase any single outlet, only the bucket as a whole.
Bias rating
NOT AI. Bias labels (Left/Center/Right) are sourced from AllSides and Media Bias/Fact Check, mapped at the publication level.
Story grouping
Headline similarity (cosine over embeddings) + manual review. Stories never merged across topics.
Source ratings

Where the Left / Center / Right labels come from.

Prism does not assign its own bias ratings. Labels are inherited from two independent media-rating organizations and re-checked quarterly:

Disagree with a specific label? Email us with the outlet name and your evidence. Every correction request is logged and replied to within 7 days.

Corrections & takedowns

If we got it wrong, we say so — in writing, with a date.

  • Mistakes in AI-generated content (Common Ground claims, Flip-the-Script rewrites): we publish a dated correction note on the story page within 24 hours of a verified report.
  • Source mis-categorization: dated note added to the source page (e.g. /sources), and re-checked against AllSides + MBFC.
  • Takedown requests from an outlet: reviewed within 72 hours. We never remove fair-use headline grouping, but we will adjust attribution language on request.
See the full corrections log →
Editorial independence

No ads, no paywall, no investors with an axe to grind.

Prism News has no advertisers, no subscription tiers, no investors with editorial influence, and runs no native promoted stories. The product is free because the cost of running it is low and the value of being trusted is enormous. If that ever changes, this page is the first place we'll say so.

Contact

For corrections, press, or just to say hi.

hello@prismnews.io