Describe your project

TruthLens: AI-Powered Fake News Detector

Project Purpose

TruthLens detects AI-edited content and digital manipulations creating misinformation for publicity. It aims to eradicate such content for public welfare and safety, serving as a crucial tool to combat fake news and protect public discourse integrity.

Impact

• News Channels: Tracks real vs. fake information, maintaining credibility.
• General Public: Distinguishes genuine from AI-manipulated content.
• Victims of AI Editing: Identifies and removes non-consensual personal media.

Technology

Gemini AI

Gemini AI, the core of TruthLens, is a cutting-edge AI system by Google that offers multimodal analysis, advanced NLP, pattern recognition, and continuous learning capabilities. These features enable comprehensive fake news detection across text, images, and potentially video content, with the ability to adapt to emerging manipulation techniques.

Infrastructure

• NLTK and scikit-learn for preprocessing
• TfidfVectorizer for feature extraction
• PassiveAggressiveClassifier for model training
• Confusion matrix for accuracy evaluation

Google News API

Fetches live news for real-time detection.

Kitware – Dive Integration

Enhances visual media analysis, suitable for dynamic social media applications.

Scope

In-scope:

• Text-based news analysis
• Major news source integration
• User-friendly interface
• Basic image analysis

Out of scope:

• Advanced video deepfake detection
• Real-time social media analysis
• Automated content removal

Future Opportunities

1. Website plugin integration
2. Browser extension
3. Social media integration
4. Mobile app development
5. API services
6. Fact-checker collaboration
7. Educational platform integration

TruthLens, leveraging Gemini AI and Kitware's Dive, aims to become indispensable in fighting digital misinformation, ensuring a safer online environment.

Challenges we ran into

AI model accuracy
Real-time processing
Multimodal analysis integration
Scalability issues
Data privacy concerns
Evolving manipulation techniques
Cross-platform compatibility
Latency in video analysis
False positive management