AI TOOL FOR COMBATING MISINFORMATION

Spot the Fake. Trust the Real.

Comment

Misinformation spreads faster than truth, especially on social media. From celebrity death hoaxes to dangerous health myths, false claims mislead people, erode trust, and sometimes cause real harm. We wanted to build a simple, AI-powered tool that helps users quickly verify claims with credible evidence before they share.

What it does

Takes in text, links, or images of a claim.

Extracts the core statement.

Cross-checks it against trusted sources & fact-check databases.

Uses AI to classify whether the claim is Supported, Refuted, Mixed, or Unknown.

Provides evidence snippets with links so users can see the truth themselves.

Outputs a confidence score and simple explanation to keep it transparent.

How we built it

Frontend: Built using AWS PartyRock for rapid prototyping with a clean UI and widget chaining.

AI Models: Leveraged Amazon Bedrock foundation models for claim extraction and reasoning.

Evidence Retrieval: Integrated fact-check APIs and curated databases into the pipeline.

Chaining: Connected multiple widgets (input → preprocess → retrieval → verdict → output) in PartyRock.

Deployment: Hosted as a PartyRock app with potential to scale into AWS services (API Gateway, DynamoDB, OpenSearch).

Challenges we ran into

Ensuring the AI doesn’t “hallucinate” and only uses verified evidence.

Handling multilingual claims (many fake posts in regional languages).

Integrating external fact-check APIs smoothly inside PartyRock.

Balancing user-friendly explanations with technical accuracy.

Accomplishments that we're proud of

Built a working prototype in less than a day using PartyRock.

Created a transparent AI flow that always cites sources.

Designed a system that can handle text, links, and even images.

Made misinformation detection accessible and easy-to-use for everyone.

What we learned

How to leverage AWS PartyRock for rapid AI app prototyping.

The importance of Retrieval-Augmented Generation (RAG) for grounding AI answers.

Fact-checking isn’t just about saying “true/false” — users need evidence they can trust.

Misinformation detection requires a human-in-the-loop approach for edge cases.

What’s next for AI Tool for Combating Misinformation

Expand multilingual support (Hindi, Telugu, Tamil, etc.).

Add image/video forensics (reverse image search, deepfake detection).

Build a browser extension so users can verify claims while browsing.

Integrate graph analysis (spot bot networks spreading the same fake news).

Transition from PartyRock prototype to a scalable AWS backend with Bedrock, OpenSearch, and Guardrails.

Built With

  • amazon-bedrock
  • and
  • built-with-aws-partyrock
  • fact-check-apis-(claimreview
  • javascript
  • politifact)
  • react
  • snopes
Share this project:

Updates