Logo Lanfrica

Fredrickighile/endphishai

Domaine:

digital infrastructurenatural language processing

Type de record:

software
Créateur:
Fre
Hôte:
AI-powered phishing detection protecting African SMEs from cyber threats. Real-time analysis. Instant alerts. Zero compromise. # EndPhishAI — Multilingual Phishing Detection Platform **Fredrick Ighile** | github.com | linkedin.com 🏆 **Top 3 Finalist — AfriHackBox Pan-African Cybersecurity Hackathon 2025** 🌐 **Live Demo** | 📡 **API** | 🎥 **Demo Video** --- ## The Problem Nobody Is Solving Every major phishing detection tool — VirusTotal, Google Safe Browsing, PhishTank — was built for English-speaking Western markets. Meanwhile, 600 million people across Africa are receiving phishing attacks in **Swahili, Hausa, Yoruba, French, and Pidgin**. These attacks target M-Pesa, MTN, GTBank, Safaricom — brands that don't exist in any Western threat database. _"Thibitisha akaunti yako ya M-Pesa haraka!"_ — this Swahili phishing SMS bypasses every existing security tool. Not EndPhishAI. --- ## Live Demo --- ## What Makes It Different | Feature | Most Tools | EndPhishAI | | ---------------------- | ---------------------------------- | ---------------------------------------------------- | | Detection approach | Reactive — checks reported threats | Proactive — analyzes behavior and structure | | Languages | English only | English, Swahili, Hausa, Yoruba, French, Pidgin | | African brand coverage | None | MTN, M-Pesa, GTBank, Airtel, Safaricom, Flutterwave | | Typosquat detection | Basic | Homoglyph matching (paypa**1**.com → PayPal) | | Subdomain spoofing | Missed | Caught (accounts.google.com.**security-alert.net**) | | Zero-day detection | No | Yes — structural analysis catches unreported threats | | Explainability | "Threat detected" | Lists exactly which signals triggered the verdict | | Input types | URLs only | URLs, SM …