Our Lab Notes

A silent cyberwar is unfolding — and your trusted two-factor authentication might no longer be enough. A new breed of phishing known as Browser-in-the-Middle (BitM) is exploiting even the most secure login mechanisms — and it’s time we pay attention. In a recent study presented at the Advanced Network Technologies and Intelligent Computing (ANTIC 2023) conference, researchers proposed a novel machine learning-based approach to detect this growing cyber threat. 📘 Study title: Phishing Detection in Browser-in-the-Middle: A Novel Empirical Approach Incorporating Machine Learning Algorithms 📌 DOI: https://doi.org/10.1007/978-3-031-64067-4_9 🌐 What Is Browser-in-the-Middle (BitM) Phishing? BitM phishing is not your everyday phishing attack. It works by placing a malicious browser "between" the victim and the legitimate website — intercepting login credentials in real-time, even when multi-factor authentication (MFA) is enabled. This effectively bypasses traditional phishing filters and authentication safeguards, making it one of the most dangerous and hard-to-detect threats in the wild. 🎯 Why It Matters: Targets MFA systems, which were previously considered secure. Mimics legitimate sessions to trick even the most tech-savvy users. Difficult to detect using traditional anti-phishing tools. 🤖 A New Defense: Machine Learning for BitM Detection While most research focuses on mitigating the effects of BitM attacks, this study takes a proactive stance — proposing a detection-first approach using machine learning algorithms trained on internally curated network packet data. 🔍 Key Insights from the Research: Custom dataset generated due to lack of public BitM-specific data. Five ML classifiers tested: SVM, MLP, Naive Bayes, Decision Tree, and Random Forest. Random Forest achieved the highest detection accuracy of 99.1%. Inspired by the high CAPEC severity rating of BitM attacks. “BitM phishing is not just another scam — it’s an intelligent and evasive cyber-weapon,” says lead researcher Md. Farhan Shahriyar. ⚠️ Why This Should Concern Everyone From banking portals to email logins, BitM phishing targets high-value services. Since most users rely on MFA as the ultimate layer of defense, this new technique undermines trust in standard security practices. If attackers can intercept live sessions, credentials, and OTPs in real-time, the implications are enormous — affecting consumers, enterprises, and even critical infrastructure. 🧠 What’s Next? From Research to Real-World Application While this approach shows high promise in experimental settings, broader deployment requires: Wider dataset expansion for training and generalization. Integration into real-time phishing detection systems. Collaboration between cybersecurity firms and academia to adopt proactive detection over reactive mitigation. The takeaway? Detection, not just prevention, is the way forward. 📝 Reference: Md. Farhan Shahriyar, Zarif Sadeque Seyam, Ahsan Ullah & Md. Nazmus Sakib (2023). Phishing Detection in Browser-in-the-Middle: A Novel Empirical Approach Incorporating Machine Learning Algorithms. Published in Advanced Network Technologies and Intelligent Computing (ANTIC 2023), CCIS volume 2093. 📖 DOI: https://doi.org/10.1007/978-3-031-64067-4_9 🔒 Final Thoughts BitM phishing is a wake-up call for cybersecurity professionals and users alike. As attackers grow smarter, so must our defenses — and that begins with understanding the threat, developing intelligent detection systems, and adapting faster than the adversaries. We may be entering a new chapter in phishing — are we ready to respond?

In today's digitally connected world, most of us view SMS messages as harmless — a quick way to get updates from friends, family, service providers, or businesses. But beneath this everyday convenience lies a rapidly growing threat: Smishing. While phishing via email has long been a known attack vector, smishing — or SMS phishing — is the new frontier for cybercriminals, and it’s hitting closer to home than many realize, especially in Bangla-speaking regions. 📉 What Is Smishing and Why Should You Worry? Smishing is a form of social engineering attack where malicious SMS messages are used to trick recipients into revealing personal data, clicking harmful links, or even transferring money. These messages often appear to come from trusted sources like banks, government agencies, or telecom operators — making them dangerously convincing. Globally, smishing attacks have surged by 328%, resulting in over $54.2 million in losses in 2019 alone. In Bangladesh and other Bangla-speaking areas, the problem is growing even faster — and it’s alarmingly under-addressed. 📱 Why Bangla Smishing Is Especially Dangerous Bangla smishing poses unique challenges: Language complexity: Many spam filters and models are not optimized for the Bangla language, especially in informal or regional variations used in texts. Lack of awareness: Users often don't recognize fake messages in Bangla due to their realistic tone and cultural context. Limited AI models: Most existing smishing detection systems are tailored for English or other major languages, leaving Bangla users exposed. 🧠 A Smarter Solution: AI-Powered Bangla Smishing Detection To combat this issue, researchers have developed a hybrid machine learning model that combines BERT (for contextual understanding of Bangla text) with character-level CNNs (for capturing subtle text patterns). This model can accurately classify Bangla SMS into three categories: Normal, Promotional, and Smishing — going beyond traditional binary classifiers. Key highlights from the study: Integrates BERT embeddings with CNN for character-level features. Employs an attention mechanism to focus on the most important parts of the message. Achieves a remarkable 98.47% accuracy in detecting smishing messages. 📝 Citation: Gazi Tanbhir, Md. Farhan Shahriyar, Khandker Shahed, Abdullah Md Raihan Chy, Md Al Adnan, Hybrid Machine Learning Model for Detecting Bangla Smishing Text Using BERT and Character-Level CNN, 2024 IEEE International Conference on Electrical and Computer Engineering (ICECE). 🔗 DOI: 10.1109/ICECE64886.2024.11024872 🔐 The Path Forward: Awareness + AI Stopping Bangla smishing requires a two-pronged approach: Public awareness: Users must learn to recognize and report suspicious SMS messages — especially those in Bangla. Advanced technology: We need AI-driven models like the one described above to be deployed by telecom operators, government agencies, and security services. Smishing is more than just a tech problem — it’s a societal threat. As mobile usage continues to rise in Bangladesh and other regions, proactive defense is essential. 💬 Final Thought Bangla smishing is no longer a niche problem — it’s a growing digital epidemic. But with smarter AI and informed users, we can build a safer mobile future for millions. If you're working in cybersecurity, mobile communications, or AI, now is the time to act.