How we try to stay honest — and why "unbiased" doesn't mean "no point of view."
UBH is not a news aggregator — and it's not a traditional newsroom. We don't sort, rank, or re-link other outlets' stories. We use a purpose-built AI agent to write original briefings from reporting across the political spectrum, and every briefing links its sources so you can verify the work yourself.
Why an AI agent? Because we believe it's the only honest way to do this at scale. A human editorial room brings human blind spots; an AI agent with the right rules, full-spectrum inputs, and transparent sourcing applies the exact same standard to every outlet, every story, every day. The setup — the rules below — is the editorial judgment. The agent executes it without favorites.
Unbiased doesn't mean having no worldview. Everyone who reports the news makes choices. What makes us different is that our choices are transparent: we show the inputs we read, we compare how every side covered the same story, we back claims with real source text, we score political bias openly, and we hold every outlet — left, right, and center — to the exact same standard. No sacred cows.
Every headline you read goes through the same pipeline. Here's what happens behind the scenes.
We pull from outlets across the entire political spectrum — left, center-left, center, center-right, and right. The goal is to read what everyone is reading, not just one tribe.
We group coverage of the same underlying event from different outlets together. That lets us see, side by side, how the same facts get framed differently.
Not everything published is news. A story has to clear a relevance and significance bar before we cover it — we filter out noise, fluff, and pure outrage bait.
We work from what outlets actually published — never invented quotes, never imagined details. If it isn't in a real source, it doesn't go in the story.
When coverage of a story is thin or one-sided, we deliberately go look for more outlets — especially from the side that's under-represented — before we publish.
If a story is only being covered by one side of the spectrum, we flag that imbalance instead of pretending the coverage is balanced when it isn't.
We don't copy, excerpt, or link-sort any outlet. Every story is an original briefing our AI agent writes from what every side reported — the sources are inputs, the briefing is the product.
Our headlines describe what actually happened — no fake suspense, no rage-baiting, no "you won't believe." Plain, accurate, and to the point.
We keep the timeline straight: what happened when, what's new versus old, and we don't recycle stale developments as if they just broke.
Every story carries an openly published political bias score from −1 to +1, so you can judge the lean for yourself.
A story isn't finished when it's written. It has to survive a second round of checks before it ever reaches you.
After a story is drafted, a separate pass re-examines it for slant — checking that the summary itself didn't quietly pick a side.
Claims are checked back against the underlying sources. If something can't be verified, it gets softened or cut — not published as fact.
If the source material is broken, empty, or can't be fetched reliably, we hold the story rather than publish something built on bad inputs.
We don't republish the same story twice. An update only goes out when there's a genuine, material change — not cosmetic rewrites.
The rules above apply identically to every outlet and every side. Nobody gets a pass for being on "your team" or ours.
We're honest that no editorial process is a blank slate. Where we apply judgment, it leans toward common-sense, fact-first principles: verifiable claims over vibes, plain language over spin, and respect for the reader's ability to make up their own mind.
That is not a partisan agenda. It's a commitment to method. When we get something wrong, the fix is the same as the standard — show the sources, check the claim, correct the record, and apply it to everyone equally.
