Media
2 items
Description
Block ads and unwanted content with AI-powered semantic rules Block unwanted content intelligently using AI-powered rules. AI AdBlocker uses advanced machine learning to detect and hide advertisements, promotional content, and other distracting elements based on semantic understanding, not just static filter lists. 🤖 THREE AI-POWERED DETECTION METHODS • Embedding-Based Rules: Fast semantic similarity matching using text embeddings • Prompt-Based Rules: Direct AI analysis for contextual content evaluation • Vision-Based Rules: Visual AI analysis to detect image-based ads and deceptive layouts 🎯 KEY FEATURES ✓ Flexible rule system with CSS selector support ✓ Semantic understanding of content meaning, not just keywords ✓ Visual blurring of blocked elements ✓ Real-time testing and debugging ✓ Multiple AI provider support (local and cloud) 💻 AI PROVIDER OPTIONS • Chrome Built-in AI (Gemini Nano) - FREE, completely private, runs on-device • Local Model Servers (LM Studio) - FREE, self-hosted • OpenAI / OpenRouter - Cloud-based (API key required) 📝 EXAMPLE RULES div[class*="ad"]:contains-meaning-embedding('advertisement') article:contains-meaning-prompt('clickbait or sensational headlines') img:contains-meaning-vision('product advertisement or sponsored content') ⚡ CHROME BUILT-IN AI SUPPORT Works with Chrome 141+ and Gemini Nano for completely free, private AI-powered blocking without any API keys or costs. The AI runs entirely on your device. ⚠️ EXPERIMENTAL DISCLAIMER This extension is experimental and under active development. Use at your own risk. It requires configuration and may need tuning for optimal results on different websites. 🔧 REQUIREMENTS • Chrome 138+ (Manifest V3) • For local AI: Chrome Built-in AI enabled OR local model server • For cloud AI: OpenAI/OpenRouter API key (optional) 🛠️ HOW IT WORKS 1. Install and configure your preferred AI provider 2. Add custom rules using the semantic syntax 3. Rules are automatically applied when pages load 4. Matched elements are visually blurred 5. Test and refine rules for better accuracy Perfect for users who want intelligent, customizable content filtering that goes beyond traditional static filter lists. For issues or questions: https://github.com/maximtop/ai-blocker-experiment or maximtop@gmail.com