• 

• 

• 

• 

• 

• 

  check the description the bar which indicates plagiarism

The Story Behind the Project: Verishield

What Inspired Me

The turning point for me happened when a close friend of mine fell victim to a highly convincing, AI-generated scam. They found an online posting for a prestigious job/internship opportunity that looked completely flawless—complete with synthetic corporate branding, realistic executive profiles, and fabricated news write-ups. My friend bought into it completely, only to realize later it was a total fraud. Seeing someone close to me get manipulated by hyper-realistic synthetic media made me realize how dangerous information voids are in real time, and I knew I had to build a tool to help people verify what they see instantly.

What I Learned

To turn this idea into a reality, I had to dive headfirst into browser ecosystems. I learned the ins and outs of how Chrome Extensions operate—understanding manifest files, background scripts, content scripts, and how to safely inject security overlays into active web pages. I also gained deep experience in prompt engineering and API orchestration, discovering how to leverage advanced language models to analyze page context on the fly without lagging the browser.

How I Built It

I built this project using Google Gemini as the core analytical engine. The extension works by actively scanning the user's current webpage, extracting questionable news text or metadata, and running it through Gemini's API. The AI evaluates the content against known lateral reading parameters and cross-references it with trusted databases to flag potential synthetic media or unverified breaking news claims before the user can share it.

The Challenges I Faced

Building the backend logic was the easy part; the real battlefield was the frontend. I quickly learned that while AI is incredibly powerful at processing data, it is notoriously bad at creating intuitive user experiences. I faced massive challenges trying to format the AI's raw output into clean, scannable UI components. The generated code frequently broke layouts or felt clunky. I had to spend countless hours manually refining the CSS, managing asynchronous states, and restructuring the UI to ensure the extension provided a smooth, non-intrusive, and beautiful experience for the everyday user.

Built With

• css3
• html5
• huggingface
• javascript
• python
• pytorch