PHISHGUARD AI
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote
PHISHGUARD AI is an AI-powered browser security assistant that helps users identify suspicious and potentially phishing links while browsing websites such as Facebook. The Chrome extension scans links visible on the current page and analyzes them using a layered security approach. A deterministic URL security engine first detects indicators such as suspicious domain structures, brand impersonation, punycode, unusual hostnames, credential-themed words, raw IP addresses, and other phishing signals. The collected evidence is then analyzed using Meta Model API. Meta's model evaluates the security signals, assigns a phishing-risk category, and generates an understandable explanation for the user. PHISHGUARD AI presents links as Low Risk, Suspicious, High Risk, or Unknown, together with supporting evidence and a risk score. The system is designed as a defensive security assistant. It does not automatically visit suspicious URLs and does not treat AI output alone as definitive proof that a link is malicious.