Inspiration
We created a chrome extension that helps users to quickly verify the credibility of text claims and detect AI-generated images without having to leave their tab or download another application.
What it does
Users can verify information with ease:
Verify Text Claims: Highlight text, right-click, and select "Verify Claim" to see a verdict, confidence level, reasoning and links to sources.
Detect Deepfakes: Right-click on an image and select "Detect Deepfake" to view the likelihood of it being AI-generated.
How we built it
Frontend:
- UI: We designed the results pop-up box using HTML, CSS, and JavaScript.
Backend:
Text Claim Verification: We integrated the sonar model using Perplexity API for text verification, using a custom prompt to generate a response with verdict, confidence level, reasoning, and sources for each claim.
Deepfake Detection: We utilised multiple AI models from HuggingFace to cover deepfakes generated with different techniques, and implemented a FastAPI backend to expose API endpoints for the Chrome extension to interact with the models. The models work best on images generated using Stable Diffusion, MidJourney V6 and Flux. The models are currently containerised with Docker and hosted on Google Cloud Run.
What's next
We plan to develop our own models and refine them over time by using user feedback to further train the model.
More about our project here.

Log in or sign up for Devpost to join the conversation.