Misinformation spreads faster than the truth and most fake news detectors are too simple to keep up. They read the text, run it through a model, and spit out a verdict. No context. No cross-checking. No real intelligence behind the answer.

Veritas ML does it differently.

The system runs every claim through three independent verification signals simultaneously. First, a Passive-Aggressive Classifier trained on TF-IDF features analyzes the text for linguistic anomalies, clickbait patterns, and vocabulary red flags. Second, a live cross-source validation engine fuzzy-matches the claim against real-time RSS feeds from BBC, CNN, Reuters, Al Jazeera, and Dawn checking whether credible outlets are reporting the same story. Third, the system queries fact-checking databases including PolitiFact to surface any existing debunks.

The result isn't just "Real" or "Fake" it's a calibrated confidence score with full transparency: which sources matched, what the fact-checkers found, and exactly why the system reached its verdict.

Running in the background, an automated ingestion pipeline fetches and classifies breaking news every 30 minutes so the system is always current, always learning from what's actually happening in the world.

The frontend delivers everything in a clean, dark glassmorphic dashboard with live search, verdict filtering, and a full audit log built with zero framework dependencies for maximum speed.