Zur Hauptnavigation wechseln Zur Suche wechseln Zum Hauptinhalt wechseln

Phishing Literacy in the Age of AI-Generated Fraud Attempts: A Quantitative Study of Consumer Knowledge, Perception and Educational Priorities

  • Markus Schendlinger

    Studienabschlussarbeit: Masterarbeit

    Abstract

    In today's digital landscape, phishing has become one of the most widespread and costly forms of cybercrime, driven in part by the continued shift of everyday consumer activities online. Phishing attacks rely heavily on psychological manipulation in order to trick individuals into disclosing sensitive information. Recent advances in artificial intelligence have further increased the effectiveness of phishing attacks. Large language models enable the creation of grammatically flawless messages, while voice cloning and deepfake technologies allow attackers to convincingly impersonate trusted individuals. As a result, many of the traditional warning signs of phishing have disappeared. This raises the question of the effectiveness of conventional awareness training.

    Despite the growing sophistication of these threats, there is currently no established framework for assessing how well consumers can recognize and respond to phishing. Existing literacy concepts address relevant aspects, but fail to account for adversarial intent, psychological exploitation and the rapid evolution of AI-driven attack methods.

    To address this gap, this thesis introduces the concept of phishing literacy. It contains three key dimensions: (1) knowledge of phishing threats, (2) awareness of persuasion and social engineering techniques and (3) knowledge of appropriate protective responses. Self-efficacy is measured separately as a complementary variable, enabling a comparison between perceived and actual competence. The framework is developed through a literature review of existing research and empirically tested using a quantitative survey of 58 German-speaking adults.

    The results support the proposed multidimensional structure and provide several key insights. Overall phishing literacy was moderate (M = 2.21 on the 0–3 summated index), with the lowest performance observed in threat knowledge (M = .67). A significant gap was identified in awareness of AI-driven phishing methods compared to traditional forms (d = 1.30), with AI-powered chatbot phishing, website clones, deepfakes and voice cloning scoring below the scale midpoint. The findings also reveal a pronounced overconfidence effect (d = 1.25), with individuals who had lower objective knowledge tending to substantially overestimate their abilities. Education level (d = 1.71) and prior training (d = 1.08) emerged as strong predictors, and older participants generally performed worse than younger ones.

    Based on these findings, the thesis recommends a sequenced approach to phishing education. First addressing overconfidence through simulation-based calibration; next closing the AI-awareness gap through micro-learning modules, deepening understanding of persuasion mechanisms through interactive exercises; and lastly, maintaining competence through ongoing training across multiple communication channels. As AI-enhanced phishing continues to erode the reliability of traditional detection strategies, building phishing literacy across all three dimensions becomes a necessary component of consumer protection in digital environments.
    Datum der Bewilligung2026
    OriginalspracheEnglisch
    Gradverleihende Hochschule
    • FH OÖ
    Betreuer/-inWolfgang Jonas Weitzl (Betreuer*in)

    Studiengang

    • Digital Business Management

    Zitieren

    '