Zypher AI

AI-powered phishing detection for Gmail. Two independent BERT models cross-check every email. Scan with one click.

Zypher AI catches phishing before you click. Paste an email, message or link, or click "Scan with Zypher" right inside Gmail, and two independently trained AI models (both fine-tuned BERT classifiers) cross-check it in seconds. HOW IT WORKS - Two models, one verdict: an Enron-trained model and a Kaggle-trained model each score the text. - When both agree you get a clear "Likely phishing" or "Looks safe" result with a confidence score. - When they disagree you get an honest "Uncertain , verify manually" instead of a confident guess. FEATURES - One-click "Scan with Zypher" button in the Gmail toolbar. Results appear as an on-page card. - Right-click any selected text on any website and choose "Scan with Zypher AI". - Paste-and-scan from the extension popup. - Desktop notification and toolbar badge (!, OK, ?) for every scan. - Free plan: 10 scans. Pro plan: 1,000 scans a month, billed monthly and cancellable any time. - Create your account inside the popup. No card needed to start. BUILT ON REAL RESEARCH The models come from the PhishGuard research project (MSc Cybersecurity, University of Derby), with about 99.5% accuracy on their test datasets. That means Zypher AI can occasionally be wrong. Treat it as a second opinion, not a guarantee, and never rely on a "safe" result alone before trusting a link or a request for money. PRIVACY Zypher AI only reads the text you choose to scan. It doesn't read your inbox in the background, doesn't sell your data, and doesn't use your scans to train models. Full policy: https://zypher-ai.netlify.app/privacypolicy.html Built by Zain Asif. Support: zaynasif59@gmail.com