• Presentation Logistics

    • Use a mic that must be passed around; maintain about six inches distance for optimal sound pickup.
    • Time limit: 7 minutes per group, with reminders for time management.
    • Questions about presentation logistics; clarify high-level questions allowed when multiple groups cover the same topic.

  • AI-Generated Phishing Email Risks and Detection Strategies

    • Significant rise in AI-generated phishing emails (1265% increase reported for late 2022 to early 2024).
    • Three research questions explored:
    1. Growth metrics of AI-generated phishing for 2020-2024.
    2. Differences in linguistic structures between AI and human phishing emails.
    3. Evolution of spam detection tools in the last decade.
    • Employed ML models: Logistic Regression, BERT, Tree-based models.
    • Notable model performance: Logistic Regression + TF-IDF achieved 98% precision, recall, and F1 scores.
    • Linguistic findings: AI emails have more correct grammar, fewer sentence lengths, and adapt tones better than human-generated emails.
    • Future work includes expanding datasets, enhancing model robustness against phishing.

  • Security and Privacy Risks in Cookie Management

    • Investigates how poor cookie practices lead to privacy risks.
    • Collected data from 7651 cookies across popular websites using simulation scripts.
    • Found compliance issues with GDPR, with 70% non-compliance observed.
    • Identified HTTPOnly and Secure flags are underused in cookie management.
    • Recommendations for better cookie practices and user privacy.

  • Email Security Protocols Exploitation

    • Studied four email providers (Gmail, Outlook, iCloud, Yahoo) regarding SPF, DKIM, and DMARC policies.
    • Conducted phishing detection analysis through simulating phishing emails.
    • Showed how spoofing vulnerabilities exist, especially under certain configurations.

  • QUIC Protocol Analysis

    • Discussed the QUIC protocol's advantages, such as improved latency and head-of-line blocking mitigation.
    • Explored DDoS and reflected amplification threats relating to QUIC configurations.
    • Found many DNS servers incorrectly implemented security, such as source address validation.

  • Browser Security and Privacy Comparative Analysis

    • Analyzed five major browsers for privacy dimensions and security threats from phishing/malware.
    • Measures against fingerprinting and phishing protection vary widely.
    • Evaluated TLS and secure protocol handling - Firefox leads in some areas.

  • Machine Learning for Intrusion Detection Systems

    • Examined how ML-based Intrusion Detection Systems (IDS) can be affected by backdoor attacks.
    • Emphasized machine unlearning to restore security in the presence of backdoor threats.
    • Focused experiments on the CIC IDS dataset, achieving notable accuracy metrics.

  • IoT Device Security Evaluation

    • Explored vulnerabilities across different categories of IoT devices with Shodan.
    • Noted self-signed certificates prevalence, exposing susceptibility to MITM attacks.
    • Recommendations for improving TLS/SSL security practices across device categories.

  • Media Streaming over QUIC

    • Investigated traffic fingerprinting for video streams over QUIC.
    • Noted patterns in average bit rates of different video qualities, with accuracy metrics reported.
    • Discussed recommendations for enhanced security features to protect against fingerprinting.

  • Final Remarks

    • Encouragement for further investigation in multiple fields discussed.
    • Reminder of the importance of security protocols in evolving digital environments.
    • Note on scheduling changes for future presentations, stressing adherence to time limits.