SafeShop AI

• 

  Showing GREEN low risk example.

• 

  Showing YELLOW medium risk example.

• 

  Showing RED high risk example.

• 

  Cover page of Safeshop AI Pitch Deck.

• 

  Screenshot of Github Repo of the project.

This project is deeply personal. I've been involved with computers my whole life, but I've also been a victim of identity theft and bank fraud. I know the feeling of helplessness and the sheer headache and heartache that follow. For years, I knew bad things were happening online, but I didn't know how to stop them.

I wanted to build a tool that could have saved me.

One of the many hats I wear happens to be a cybersecurity researcher, so I knew the core problem: attackers don't just hack, they hunt for easy targets. They share lists of sites with non-obvious flaws, like no 3D Secure. Consumers can't see these red flags. When I saw the Google Chrome AI Challenge, I knew this was my chance to build the solution: SafeShop AI. (pretty straightforward name for a powerful tool)

What It Does

SafeShop AI is a Chrome Extension co-pilot. When you land on a checkout page, you click its icon to run an instant analysis.

1. AI Analysis: It uses the Chrome Prompt API to feed my research directly to an on-device AI model, which scans for the exact red flags attackers exploit (CVV fields, 3D Secure, etc.).

2. AI Summary: It pipes that technical report into the Summarizer API to generate a simple, color-coded "Green," "Yellow," or "Red" risk assessment, with a clear "Why" and "Advice."

The entire process is 100% private. Because it uses the new built-in AI APIs, no sensitive checkout data ever leaves your browser.

The Challenge: Building with an AI Partner

Building this was a challenge in itself. The built-in AI APIs are brand new and experimental. My development process was a 'vibe coding' session with an AI partner, and we hit numerous hurdles. We debugged ghost errors in Chrome Canary, battled caching issues, and worked through model unavailability. But this challenge proved the concept: if a human researcher and an AI can team up to build a solution in a few days, imagine what this tech can do for millions of users. We built a 'mock' framework to ensure the demo is stable, but the AI logic is the real, battle-tested engine.

The Future

This MVP is the foundation for a full B2C and B2B security suite. My goal is to create a tool that not only warns you but, in the future, can proactively alert you—the exact feature I wish I'd had.

Built With

• api
• chrome.extension
• css3
• cybersecurity
• ethical-hacking
• html5
• javascript
• nano
• promptai
• summarizer