ScamShield combines several detection layers. Here is what powers each verdict:
A built-in, offline rule engine covering 9 global scam categories — fake delivery, bank phishing, government impersonation, fake legal threats, investment fraud, romance scams, tech-support scams, lottery/prize scams and job-offer scams — plus urgency-pressure and sensitive-data-request signals. Developed in-house; runs entirely on your device.
Heuristics for phishing URLs: IP-literal links, suspicious/free domain extensions, brand impersonation in subdomains, URL shorteners and insecure HTTP. Runs offline.
Pattern checks for known fraud number prefixes, high-risk international country codes and abnormal number formats. Runs offline.
Optional online reputation lookups for links and file hashes against 70+ antivirus engines. Used only when you start a scan. ScamShield is not affiliated with or endorsed by VirusTotal.
Optional deep content analysis in the AI Assistant. Text is sent only when you explicitly submit it. ScamShield is not affiliated with or endorsed by OpenAI.
Risk levels (Safe / Low / Medium / High / Critical) are statistical estimates produced by the layers above. They are guidance, not certainty — see our Terms of Service.