Protection Against Semantic Social Engineering Attacks

Abstract

Phishing, drive-by downloads, file and multimedia masquerading, domain typosquatting, malvertising and other semantic social engineering attacks aim to deceive the user rather than exploit a technical flaw to breach a system’s security. We start with a chronological overview to illustrate the growing prevalence of such attacks from their early inception 30 years ago, and identify key milestones and indicative trends which have established them as primary weapons of choice for hackers, cyber-criminals and state actors today. To demonstrate the scale and widespread nature of the threat space, we identify over 35 individually recognised types of semantic attack, existing within and cross-contaminating between a vast range of different computer platforms and user interfaces. Their extreme diversity and the little to no technical traces they leave make them particularly difficult to protect against. Technical protection systems typically focus on a single attack type on a single platform type rather than the wider landscape of deception-based attacks. To address this issue, we discuss three high-level defense approaches for preemptive and proactive protection, including adopting the semantic attack killchain concept which simplifies targeted defense; principles for preemptive and proactive protection for passive threats; and platform based defense-in-depth lifecycle designed to harness technical and non-technical defense capabilities of platform providers and their user base. Here, the human-as-a-security-sensor paradigm can prove particularly useful by leveraging the collective natural ability of users themselves in detecting deception attempts against them.