BaitBreaker - Fighting Clickbait with instant title answers

Inspiration

I was frustrated by the overwhelming amount of clickbait content online that wastes time and manipulates readers with misleading headlines. I wanted to create a solution that gives users instant answers without requiring them to click through, combining intelligent detection with on-device AI to provide privacy-friendly summaries directly in the browser.

What it does

BaitBreaker is a Chrome extension that automatically detects clickbait links on any webpage and marks them with a distinctive [B] badge. When you hover over a marked link, BaitBreaker displays a concise summary extracted from the target article, with all AI processing happening entirely on-device using Chrome's built-in Gemini Nano AI model.

How I built it

I built BaitBreaker as a Chrome extension with a three-layer architecture: a content script that scans web pages, an in-page script that accesses Chrome's built-in AI APIs (Prompt API for classification and Summarizer API for summaries), and a service worker that coordinates components and manages caching. I also implemented a regex-based detection mode as a fallback for performance.

Challenges I ran into

The biggest challenge was that it's sometimes genuinely difficult to accurately detect clickbait. Many headlines use subtle techniques or clever wording, making it tricky to distinguish between legitimate articles and clickbait with simple patterns or even with AI models.

Accomplishments that I'm proud of

I'm particularly proud of building a fully privacy-preserving solution that processes all AI operations on-device without sending any user data to external servers. Successfully integrating Chrome's cutting-edge built-in AI capabilities was a significant technical achievement. The extension provides instant value with zero setup—users just install it and immediately see clickbait indicators, with summaries preloading in the background seamlessly.

What I learned

I gained insights into Chrome's built-in AI capabilities and learned how to work with on-device machine learning models, understanding the trade-offs between on-device processing versus cloud-based solutions.

What's next for BaitBreaker

I'm planning to expand the regex detection patterns to catch more clickbait variations and improve the AI classification accuracy by fine-tuning prompts and confidence thresholds.

Built With

• chrome
• javascript
• npm