This innovative AI browser extension utilizes a retrieval-augmented large language model (LLM) to identify low-credibility posts on Facebook and X and generate thoughtful responses aimed at bridging political divides. By referencing a reliable news database updated every two hours, the extension ensures your replies are informed and relevant.
- Credibility Detection: Automatically highlights posts with low credibility.
- Informed Responses: Generate responses based on the latest news and data to foster meaningful conversations.
- Reliable Sources: Continuously updated news database every two hours to ensure accuracy.
- Political Perspectives: Automatically detects political posts by keyword search and generates side-by-side Democrat and Republican perspectives to show different viewpoints when clicked on "Review" button.
- Multi-Platform Support: Works on Facebook and X (Twitter).
- Beta Version: This extension is currently in development and is being continuously improved.
- Verification Required: Large Language Models (LLMs) can occasionally produce errors. Always verify critical information before sharing.
- Current Model: The extension currently uses OpenAI’s Web Search endpoint and GPT-4.1 model for comment generation. Users need to input their own OpenAI API keys, to generate the responses.
- Future Plans: We are actively working on integrating other OpenAI models and open-source alternatives as funding allows.
- Currently the middleware is used for only the store the response from users.
The data is stored in a google excel sheet, It with the osome.developer@gmail.com.
Please refer here to update the platformConfig.js if there is change happened to Facebook HTML elements.
When a post contains political keywords (defined in chrome_extension/config/keywords.json) and has no external URLs, the extension generates two AI perspectives:
- Democrat Perspective: Analysis from a liberal viewpoint
- Republican Perspective: Analysis from a conservative viewpoint