91% of cyberattacks start with a phishing email — but spam filters only catch bad links and known malicious domains. Modern scams use psychological manipulation instead: artificial urgency, authority impersonation, emotional pressure, and forced secrecy. Non-technical users can't recognize these tactics.
PhishLighter uses AI to detect the psychology behind social engineering — not just the technical indicators. Paste any email or message, and PhishLighter instantly highlights the exact manipulative phrases in the text with color-coded severity. Hover over any highlight to read a plain-English explanation of why that language is dangerous and what psychological trick is being used.
It detects 10 categories of manipulation tactics including artificial urgency, authority impersonation, fear/threats, reward baiting, forced secrecy, and technical deception. The tool includes 5 built-in sample emails demonstrating real-world attack types (CEO fraud, credential phishing, package delivery scam, crypto investment scam) plus a legitimate email to prove no false positives.
Tutorial is in the github repo.
Built With
- groq-api-(llama-3.3-70b)
- next.js-14
- ollama
- react-18
- tailwind-css-3
Log in or sign up for Devpost to join the conversation.