Sulla Sicurezza degli Ambienti Sensibili alla Privacy: Un'Analisi Approfondita delle Minacce Cibernetiche Digitali e Fisiche

In an era where personal data increasingly represents the new gold, the pervasive threat of data leaks and exfiltrations significantly jeopardizes individual privacy, corporate integrity, and national security. Every day, individuals and organizations face potential cyber threats that extend beyond the digital realm; contrary to common belief, these attacks can also manifest in the physical world, infiltrating our daily routines and environments through unsuspected channels. This dissertation examines the issue of privacy preservation through two distinct lenses: (i) the security implications associated with digital documents and (ii) the study of proximity-based cyber threats. The first part of this dissertation addresses digital documents as a potential attack vector, presenting an autonomous malware detection system specifically designed for MS Office documents. Additionally, it introduces RedactBuster, the first entity recognition system capable of classifying redacted text based on the contextual meaning of the surrounding sentences. The second part of the dissertation addresses two distinct proximity attacks that may occur in a shared physical working environment. BLUFADER is an innovative countermeasure against lunchtime attacks by proposing a privacy-aware continuous authentication system. Furthermore, this section includes a novel Side-Channel Attack (SCA) that infers user actions solely by leveraging the noise generated by the operation of a computer mouse.