The latest articles on Forem by Naweid Hassan (@naweid_hassan_d7d03584c2e). https://forem.com/naweid_hassan_d7d03584c2e https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3850287%2Fd698d39c-3196-423e-8dff-162741cda867.png https://forem.com/naweid_hassan_d7d03584c2e en Naweid Hassan Mon, 30 Mar 2026 01:44:50 +0000 https://forem.com/naweid_hassan_d7d03584c2e/i-built-a-chrome-extension-that-fact-checks-ai-outputs-heres-how-it-works-d7h https://forem.com/naweid_hassan_d7d03584c2e/i-built-a-chrome-extension-that-fact-checks-ai-outputs-heres-how-it-works-d7h <p>About 27% of AI outputs contain fabricated claims. I kept running into this problem — ChatGPT confidently citing studies that don't exist, Gemini inventing statistics, Claude hallucinating facts. So I built a tool to catch it.</p> <h2> What is Aretify? </h2> <p><a href="https://www.aretify.com" rel="noopener noreferrer">Aretify</a> is a Chrome extension that adds a "Verify" button to AI responses on ChatGPT, Claude, and Gemini. One click extracts every factual claim and checks it against real sources.</p> <h2> The Tech Stack </h2> <ul> <li> <strong>Frontend:</strong> Next.js on Vercel</li> <li> <strong>Backend:</strong> FastAPI on Railway</li> <li> <strong>Database:</strong> Supabase (PostgreSQL)</li> <li> <strong>Cache:</strong> Upstash Redis</li> <li> <strong>Extension:</strong> Chrome Manifest V3, side panel architecture</li> </ul> <h2> How Verification Works </h2> <p>When you click "Verify," here's what happens:</p> <ol> <li><p><strong>Claim Extraction</strong> — The AI response is sent to Groq for fast claim extraction. Each individual factual claim is isolated.</p></li> <li><p><strong>Evidence Search</strong> — Each claim is searched against 15+ evidence sources simultaneously: Semantic Scholar, PubMed, OpenAlex, Crossref, Europe PMC, Wikipedia, Wikidata, Google Fact Check, GDELT, NewsAPI, Brave Search, SerpAPI, and more.</p></li> <li><p><strong>Scoring</strong> — Claims are scored based on how well they match evidence. Each gets a status: Verified, Partial Match, or Unverified. The overall response gets an AretifyScore from 0-100.</p></li> <li><p><strong>Ethics Analysis</strong> — This is the part I'm most proud of. My father is a philosophy professor, and he developed a 35-page framework covering 8 philosophical traditions (utilitarian, deontological, virtue ethics, care ethics, and more). Every verification runs through this ethics engine to evaluate the ethical implications of AI outputs — not just accuracy.</p></li> </ol> <h2> What I Learned Building This </h2> <p><strong>API redundancy matters.</strong> With 15+ evidence sources, some are always down. I implemented circuit breakers for unreliable APIs (Tavily, GDELT) so the system degrades gracefully instead of failing.</p> <p><strong>Chrome Extension development is painful.</strong> Manifest V3 restrictions, CSP issues across different AI platforms, and the fact that <code>.ico</code> files cause context invalidation in MV3 (Chrome requires <code>.png</code>) — these were hours of debugging.</p> <p><strong>Canonical URLs matter from day one.</strong> I launched with inconsistent www vs non-www URLs and Google was confused about which pages to index. Fix this before you launch, not after.</p> <h2> Try It </h2> <p>The extension is free — 10 verifications per day, no credit card required.</p> <ul> <li>Website: <a href="https://www.aretify.com" rel="noopener noreferrer">aretify.com</a> </li> <li>Chrome Web Store: <a href="https://chromewebstore.google.com/detail/aretify-ai-output-verifie/fepnnhblfpfcoecdnmcccfjajfiijpmk" rel="noopener noreferrer">Install extension</a> </li> </ul> <p>I'd love feedback from the dev community — especially on the verification pipeline and extension UX. What would you want from a tool like this?</p> ai webdev programming productivity Naweid Hassan Mon, 30 Mar 2026 01:32:12 +0000 https://forem.com/naweid_hassan_d7d03584c2e/i-built-a-chrome-extension-that-fact-checks-ai-outputs-heres-how-it-works-49dp https://forem.com/naweid_hassan_d7d03584c2e/i-built-a-chrome-extension-that-fact-checks-ai-outputs-heres-how-it-works-49dp <p>About 27% of AI outputs contain fabricated claims. I kept running into this problem — ChatGPT confidently citing studies that don't exist, Gemini inventing statistics, Claude hallucinating facts. So I built a tool to catch it.</p> <h2> What is Aretify? </h2> <p><a href="https://www.aretify.com" rel="noopener noreferrer">Aretify</a> is a Chrome extension that adds a "Verify" button to AI responses on ChatGPT, Claude, and Gemini. One click extracts every factual claim and checks it against real sources.</p> <h2> The Tech Stack </h2> <ul> <li> <strong>Frontend:</strong> Next.js on Vercel</li> <li> <strong>Backend:</strong> FastAPI on Railway</li> <li> <strong>Database:</strong> Supabase (PostgreSQL)</li> <li> <strong>Cache:</strong> Upstash Redis</li> <li> <strong>Extension:</strong> Chrome Manifest V3, side panel architecture</li> </ul> <h2> How Verification Works </h2> <p>When you click "Verify," here's what happens:</p> <ol> <li><p><strong>Claim Extraction</strong> — The AI response is sent to Groq for fast claim extraction. Each individual factual claim is isolated.</p></li> <li><p><strong>Evidence Search</strong> — Each claim is searched against 15+ evidence sources simultaneously: Semantic Scholar, PubMed, OpenAlex, Crossref, Europe PMC, Wikipedia, Wikidata, Google Fact Check, GDELT, NewsAPI, Brave Search, SerpAPI, and more.</p></li> <li><p><strong>Scoring</strong> — Claims are scored based on how well they match evidence. Each gets a status: Verified, Partial Match, or Unverified. The overall response gets an AretifyScore from 0-100.</p></li> <li><p><strong>Ethics Analysis</strong> — This is the part I'm most proud of. My father is a philosophy professor, and he developed a 35-page framework covering 8 philosophical traditions (utilitarian, deontological, virtue ethics, care ethics, and more). Every verification runs through this ethics engine to evaluate the ethical implications of AI outputs — not just accuracy.</p></li> </ol> <h2> What I Learned Building This </h2> <p><strong>API redundancy matters.</strong> With 15+ evidence sources, some are always down. I implemented circuit breakers for unreliable APIs (Tavily, GDELT) so the system degrades gracefully instead of failing.</p> <p><strong>Chrome Extension development is painful.</strong> Manifest V3 restrictions, CSP issues across different AI platforms, and the fact that <code>.ico</code> files cause context invalidation in MV3 (Chrome requires <code>.png</code>) — these were hours of debugging.</p> <p><strong>Canonical URLs matter from day one.</strong> I launched with inconsistent www vs non-www URLs and Google was confused about which pages to index. Fix this before you launch, not after.</p> <h2> Try It </h2> <p>The extension is free — 10 verifications per day, no credit card required.</p> <ul> <li>Website: <a href="https://www.aretify.com" rel="noopener noreferrer">aretify.com</a> </li> <li>Chrome Web Store: <a href="https://chromewebstore.google.com/detail/aretify-ai-output-verifie/fepnnhblfpfcoecdnmcccfjajfiijpmk" rel="noopener noreferrer">Install extension</a> </li> </ul> <p>I'd love feedback from the dev community — especially on the verification pipeline and extension UX. What would you want from a tool like this?</p> ai webdev opensource productivity