{"slug": "i-built-termsight-an-ai-that-reads-the-privacy-policy-so-you-don-t-have-to", "title": "I Built Termsight: An AI That Reads the Privacy Policy So You Don't Have To.", "summary": "A developer built Termsight, an open-source tool that analyzes privacy policies and terms of service and explains them in plain English, covering data collection, third-party sharing, retention, and user rights, with each finding linked to the exact source section. The tool runs locally on the open-weight Qwen2.5-1.5B-Instruct model with a Streamlit interface, so policy text is never sent to a commercial AI API. The developer split the analysis into four focused prompts rather than one combined prompt, which improved answer quality and prevented the model from returning bare citations instead of explanations.", "body_md": "What I Built\n\nTermsight is an AI tool that reads privacy policies and terms of service\n\nand explains them in plain English, so you don't have to.\n\nPaste in a policy (or a link to one), and Termsight breaks it down into\n\nfour things people actually care about: what information is collected,\n\nhow it may be shared with third parties, how long it's retained, and\n\nwhat rights you might be giving up. Every finding links back to the\n\nexact section of the original document it came from — and if the\n\npolicy doesn't clearly answer something, Termsight says so instead of\n\nguessing.\n\nThe lazy payoff: nobody has to actually read pages of legal text\n\nanymore just to know what they're agreeing to. Anyone who's ever\n\nclicked \"Accept\" without reading gets to be a little lazier and\n\nbetter informed at the same time.\n\n[https://lumping-playable-sauciness.ngrok-free.dev/](https://lumping-playable-sauciness.ngrok-free.dev/)\n\nGitHub: [https://github.com/Ayomiposi20/termsight](https://github.com/Ayomiposi20/termsight)\n\nTermsight runs on Qwen2.5-1.5B-Instruct, an open-weight language\n\nmodel, loaded and run locally rather than called through a commercial\n\nAI API. The interface is built with Streamlit.\n\nThe pipeline:\n\nI deliberately split the analysis into four separate, focused prompts\n\nrather than one big prompt asking everything at once with a smaller\n\nmodel, this made a real difference in answer quality and prevented the\n\nmodel from just dumping bare citations instead of actual explanations.\n\nPrivacy policies are sensitive by nature pasting one into a\n\ncommercial AI service means trusting yet another third party with that\n\ntext, which is a strange thing to do for a tool whose whole purpose is\n\nhelping you protect your privacy.\n\nBecause Termsight runs on an open-weight model instead of a closed API,\n\nthe policy text is never sent to a third-party AI service. That's not\n\njust a technical detail it's the actual point. A tool meant to help\n\nyou understand what you're agreeing to shouldn't require agreeing to\n\nsomething new (sending your data elsewhere) just to use it. Open models\n\nmake that kind of honest, self-contained tool possible in a way a\n\nclosed API fundamentally can't.", "url": "https://wpnews.pro/news/i-built-termsight-an-ai-that-reads-the-privacy-policy-so-you-don-t-have-to", "canonical_source": "https://dev.to/ayomiposi20/i-built-termsight-an-ai-that-reads-the-privacy-policy-so-you-dont-have-to-7pa", "published_at": "2026-10-25 16:09:50+00:00", "updated_at": "2026-10-25 16:13:06.578011+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-tools", "ai-products"], "entities": ["Termsight", "Qwen2.5-1.5B-Instruct", "Streamlit", "GitHub"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/i-built-termsight-an-ai-that-reads-the-privacy-policy-so-you-don-t-have-to", "markdown": "https://wpnews.pro/news/i-built-termsight-an-ai-that-reads-the-privacy-policy-so-you-don-t-have-to.md", "text": "https://wpnews.pro/news/i-built-termsight-an-ai-that-reads-the-privacy-policy-so-you-don-t-have-to.txt", "jsonld": "https://wpnews.pro/news/i-built-termsight-an-ai-that-reads-the-privacy-policy-so-you-don-t-have-to.jsonld"}}