Inspiration:

I kept seeing friends and locals in Poland lose money to fake listings on OLX and Otodom fake landlords, duplicate ads, and scammers asking for deposits. There was no simple way to detect these scams early. That frustration sparked FraudBlok, a lightweight, privacy-first Chrome extension that warns users before they get tricked.

What it does

FraudBlok scans every OLX or Otodom listing you visit and uses AI-based pattern recognition to detect potential fraud indicators such as:

Unrealistic prices Suspicious seller names or emails Scam-like language patterns Mismatched contact information

It alerts you instantly if a listing looks risky without collecting or sharing any personal data.

How we built it

Frontend: Chrome Extension built with JavaScript, HTML, CSS Backend logic: Lightweight AI model for text-pattern detection (fine-tuned locally for Polish marketplace listings) Infrastructure: Static hosting on GitHub Pages for landing site + Chrome Web Store for distribution Privacy: All detection runs locally inside the browser no external API calls or user tracking Google Gemini Flash as the model

Challenges we ran into

Training the detection logic on realistic scam samples while keeping it light enough to run in-browser Ensuring accurate detection in Polish + English listings Balancing performance vs. privacy — making sure the extension is fast but never leaks user data Navigating Chrome Web Store publishing and automated review

Accomplishments that we're proud of

Live public release on Chrome Web Store link Fully functional AI detector that runs locally Growing community feedback from Polish OLX users 100% open-source transparency with hosted demo site: fraudblok.github.io

What we learned

Real-world fraud detection isn’t just about data — it’s about context and behavior Browser extensions can be powerful AI tools if designed responsibly Importance of clear UX in delivering trust-based alerts Privacy-first AI design builds long-term credibility

What's next for FraudBlocker

Expand support to global marketplaces (e.g., Facebook Marketplace, eBay, Gumtree) Add real-time scam report sharing and crowd-verification Integrate lightweight LLM-based classifiers for adaptive fraud patterns Launch community “verified seller” badges Explore partnerships with consumer protection agencies in Poland and EU

Share this project:

Updates