Can You Tell What's Real Anymore?

Every 3 seconds, a fake news article is shared online. We built sophisticated AI to fight back and restore digital trust.

0 Fake Articles Identified

Try the Detector See How It Works

The Misinformation Crisis Is Real

6x Faster

Fake news spreads 6x faster than real news across social networks (MIT Study).

75% Struggle

Three-quarters of adults are unable to reliably distinguish fake news from truth.

$78B Loss

Misinformation causes over $78 billion in economic damage globally every single year.

In an era of deepfakes and mass-produced content, misinformation isn't just an annoyance — it's a weapon. It compromises public health, manipulates democratic elections, and polarizes entire societies before the facts even emerge.

It Started With a Message

"URGENT: Don't open your windows tonight." A simple text, forwarded thousands of times in hours. By morning, a city was in panic. There was no emergency — just a ghost story designed to travel.

We watched how one lie, packaged correctly, could travel around the globe before the truth could even finish its first cup of coffee. That's why we created TruthGuard.

Step 1: The Fabrication

A sensationalist headline is generated by a bot or malicious actor, designed to trigger maximum emotional response.

Step 2: The Echo Chamber

Algorithms recognize the high engagement and push it to thousands of like-minded users who share it without verifying.

Step 3: The Impact

Real-world consequences occur — from store closures to medical misinformation — based on a document that never existed.

Meet Your AI Fact-Checker

We use industry-standard Machine Learning to scan patterns humans miss.

Paste Article

Simply copy and paste raw text from any news source, social media post, or messaging app.

AI Analysis

Our model analyzes over 5,000 linguistic features, sentiment markers, and structural patterns.

Instant Verdict

Get a REAL or FAKE label with a precise confidence score in less than 200 milliseconds.

Powered by Real Machine Learning

NLP Pipeline

Porter Stemming and Tokenization cleanup for raw text.

TF-IDF Vectorics

Converting words into math using Term Frequency weighting.

Logistic Regression

Advanced classification engine for ultra-fast inference.

Dual Mode

Switch between Local Models and LLM-powered API logic.

Performance Benchmark

Model Accuracy 98.7%

Validated on 44,898 articles from the official Kaggle dataset.

Input

NLP

MODEL

Result

The Numbers Don't Lie

Metric	Logistic Regression	Naive Bayes
Accuracy	98.7%	94.1%
F1 Score	0.987	0.938
Precision	98.2%	93.6%
Speed	Ultra Fast	Fast

4632True Positive

78False Positive

34False Negative

4236True Negative

Confusion Matrix Data (Testing Set)

"I use this every time I see a questionable forward on my family WhatsApp groups. It's incredibly fast."

— John Doe

"The AI reasoning feature in API mode is a game changer for understanding why news is fake."

— Alice Smith

"A perfect demonstration of NLP projects for engineering finals. Clean code and reliable metrics."

— Robert King

Built by Students, Powered by AI

Lead Dev

ML Engineer

Backend

Python & Flask

Frontend

UI/UX Designer

Final Year Project — Computer Science Department, 2024