Most AI news doesn’t make it here. That’s the point.
There’s a person behind this. Every featured story is chosen by a human editor who reads the brief, follows the sources, and decides what actually matters that day. The pipeline handles the volume; the judgment is mine.
A curated list of RSS feeds from labs, regulators, maintainers, and research venues. For each item, we fetch the full article text and read it end-to-end. No wire aggregators, no press-release laundering — if it isn't in the primary record, it doesn't count toward the score.
Every brief is original writing. We synthesise what the source says; we don't paraphrase it sentence-by-sentence. The source is cited and linked. If the underlying claim isn't falsifiable, the trust score reflects that.
Most stories don't make it. AI generates enormous noise every day — funding announcements dressed as breakthroughs, opinion pieces dressed as research, press releases dressed as news. Nothing publishes below a 50/100 trust score. When nothing clears, nothing sends. Skip days are a feature, not an embarrassment.
Named sources, multiple outlets, falsifiable claims. Take it at face value.
Real story, but details may be unconfirmed — e.g. a paywalled article scored only on its headline, or a scoop with no official statement yet. Worth reading; apply normal scepticism.
Speculative, opinion-heavy, or single anonymous source. Held back automatically.