TruthUnfiltered

Global News Network | Amplify The Silenced | Question The Powerfull

Comment

TruthUnfiltered

Amplify the Silenced. Question the Powerful.

Inspiration

News today is shaped by the loudest and most powerful voices. Corporate media, state-backed narratives, and algorithmic amplification generate overwhelming noise, while independent journalists, marginalized communities, and Global South perspectives are often silenced.

Instead of asking how to show “both sides,” we asked a harder question:

What if technology actively corrected power imbalances in information?

TruthUnfiltered was inspired by the belief that media bias is not accidental — it is structural — and meaningful bias mitigation requires amplifying voices that are systematically excluded.

What it does

TruthUnfiltered is an AI-powered news platform that mitigates knowledge bias and cognitive bias by filtering power-backed noise and amplifying silenced perspectives.

It:

  • Surfaces underreported stories and marginalized voices
  • Compares dominant narratives with independent reporting side by side
  • Exposes framing techniques, ownership, funding, and conflicts of interest
  • Breaks algorithmic filter bubbles
  • Translates non-English journalism to reduce information inequality

Rather than pretending neutrality, TruthUnfiltered levels the playing field.

How we built it

TruthUnfiltered was built as a full-stack AI system.

Backend

  • Python FastAPI
  • LangChain for AI agent orchestration
  • Gemini 2.0 Flash for narrative analysis
  • Tavily for real-time web search
  • Pydantic for structured data validation

The AI agent:

  1. Identifies major news stories by topic or location
  2. Finds dominant and opposing narratives
  3. Actively searches for independent and marginalized sources
  4. Analyzes ownership, funding, and framing
  5. Separates shared facts from interpretation

Frontend

  • Next.js web interface
  • Side-by-side narrative comparison
  • Clear labeling of power, framing, and conflicts of interest
  • Bias made visible and understandable for non-technical users

Challenges we ran into

  • Defining bias without oversimplifying a complex concept
  • Identifying truly independent sources in noisy media ecosystems
  • Balancing transparency with usability in the interface
  • Building a complete system within hackathon time limits

Each challenge directly influenced our design and technical decisions.

Accomplishments that we're proud of

  • Building a fully functional AI agent, not just a concept
  • Intentionally amplifying silenced voices
  • Making bias visible and actionable
  • Addressing both systemic and cognitive bias
  • Delivering a polished and scalable solution

What we learned

  • Bias mitigation is as much a design problem as a technical one
  • Neutral systems often reinforce existing power structures
  • Transparency builds trust more effectively than objectivity claims
  • Information access is critical to sustainability and equity
  • Small interface decisions can shape how truth is perceived

What's next for TruthUnfiltered

  • Expanding coverage of Global South and conflict-zone journalism
  • Creating revenue-sharing models for independent journalists
  • Adding community-driven verification and feedback
  • Improving multilingual and cultural bias detection
  • Partnering with journalists, researchers, and sustainability organizations `

Built With

Share this project:

Updates