If you get your news from Google News, Apple News, or Bing News, a recent audit suggests you're getting a narrower version of the news cycle than you think.
In March and April 2026, AllSides published a series of audits measuring which outlets these aggregators surface and in what proportions. The findings are consistent enough to notice. Google News's top stories drew from sources rated 73% left-leaning. Apple News displayed 50% left-leaning sources and 2% right-leaning. Bing News landed at 72% left-leaning. SmartNews, NewsBreak, and Drudge Report all skewed in various directions with the same structural result: the average reader's feed is not a cross-section of the news ecosystem.
That is not a political complaint. It is a description of what an algorithm does when it optimizes for engagement, freshness, or click-through rates using training data drawn from an uneven pool of outlets. Whatever ratio you see when you open a news app is a ratio the algorithm selected. You did not.
What a News Aggregator Is Actually Doing
Aggregators pull headlines from hundreds or thousands of publishers and rank them for each user. The ranking model weighs factors like recency, publisher authority, topical relevance, and implicit signals from what similar users clicked. On paper this is neutral. In practice, publisher authority is a trailing metric — outlets that historically generated more clicks get boosted in future rankings — and the outlets that generated those historical clicks reflect whoever was online and clicking five or ten years ago.
The result is that a story published by two outlets at the same time, on the same topic, with the same facts, can appear in radically different slots depending on which outlet published it. And a story published by only one side of the ideological spectrum may not appear at all.
Why the Split Matters
The word to pay attention to here is not bias. It is selection. A 73%-to-single-digit ratio does not mean every headline is spun. It means you are more likely to see the stories that left-leaning outlets chose to cover and less likely to see the stories that right-leaning outlets chose to cover — or the inverse, depending on the aggregator. Coverage decisions upstream become visibility decisions downstream.
This is the mechanism that produces what we call blindspots. A blindspot is not a distortion. It is an absence — a story one side's ecosystem ignored while the other side covered it heavily. We walked through the concept in more depth in our earlier piece on what a news blindspot is and how to find yours, and aggregator skew is where the mechanic shows up in everyday reading.
What This Looks Like in Practice
This week's Signal/noise dataset has several concrete examples of the pattern.
A U.S. Army request to halt production on the M109 Paladin artillery line — a significant procurement shift driven by lessons from the Ukraine war — was covered by Breaking Defense and republished by RealClearDefense. No left-leaning outlet touched it. Readers whose aggregator leaned left would have seen nothing about a decades-old ground-fires program being wound down. Our story page on the Paladin production halt request lays out what the defense press reported and what the rest of the coverage landscape left untouched.
The mirror image of that pattern also showed up this week. NPR and the New York Times covered the Trump administration's immigration crackdown in detail, including a U.S. citizen infant separated from its deported mother and a documented pattern of Salvadorans disappearing into CECOT prison after arrival. No right-leaning outlet in the coverage set picked up either angle. Readers whose aggregator leaned right — or whose feed was dominated by right-leaning publisher authority — saw the enforcement action but not the downstream human-rights reporting. The full breakdown is on our immigration crackdown story page.
A New York Times profile of Samuel Samson, the 27-year-old Trump administration diplomat driving the break with European allies, was covered by no right-leaning outlet. A preliminary federal approval for a 250-foot triumphal arch on the National Mall, advanced despite overwhelmingly negative public comment, appeared in CNN and the New York Times — and nowhere on the right. Each of these has obvious political salience to both sides. Each landed in only one half of the aggregator ecosystem.
What Algorithmic Curation Can't Show You
The structural issue is this: aggregators can only surface what publishers produce. If one side's outlets do not run a story, no ranking algorithm in the world can surface it to their readers — because the URL does not exist in the index.
This is where outlet-level bias ratings (AllSides, Media Bias/Fact Check, Ad Fontes) run into a ceiling. A ratings chart tells you where an outlet sits on a spectrum. It does not tell you which stories that outlet chose not to cover. The rating is a property of the publisher; the blindspot is a property of the story. Our guide to detecting media bias and our breakdown of why source count matters more than any single article hit the same underlying point from different angles: seeing what one side is not saying requires comparing full coverage sets, not individual outlets.
Diversifying Your Feed Without Doubling Your Reading Time
A few practical habits close most of the gap.
First, cross-check any story that feels politically charged against an aggregator with a different lean. If your primary feed is Apple News, try Ground News, Drudge, or a story-level comparison tool once a day. Our roundup of five tools that let you compare news sources side by side covers the main options.
Second, pay attention to what is absent. If a headline asserts that "everyone is talking about" something, ask whether that is true across the full spectrum or only in the outlets your feed prefers.
Third, read story-level coverage comparisons. Seeing that CNN called a ceasefire "holding" while Fox called it "Trump's historic day" is the kind of information no outlet-level bias chart can give you. That is why we built story-level framing comparisons into Signal/noise in the first place.
Algorithms are not going to stop filtering. The ratio is not going to rebalance on its own. But knowing the filter is there is the first step toward reading past it.