19th Annual eCrime Symposium Announces Accepted Papers

CAMBRIDGE, MA, UNITED STATES, October 22, 2026 /CourierPR/ -- The managing chairs of APWG’s 2026 Symposium on Electronic Crime Research (eCrime) today announced 14 accepted papers selected from more than 40 submissions to this year's APWG eCrime Symposium. These papers will be presented at eCrime 2026 Lisboa, alongside sessions featuring industry, law enforcement, NGOs, and multilateral organizations. The accepted papers will be published in IEEE Xplore, marking eCrime’s 19th year as an IEEE conference.

The accepted papers span a wide range of topics, including phishing infrastructure, criminal economies, social engineering, human behavior, information manipulation, and AI-driven defenses. Together, they highlight the need for interdisciplinary research, shared cybercrime event ontologies, interoperable intelligence, and coordinated international institutions to address the increasingly complex criminal ecosystems.

The accepted papers include:

1. This paper demonstrates the limitations of static detection methods, showing that F1-scores fall to 0.326 when dynamic features are removed.

2. A six-month comparative evaluation of Google Safe Browsing and competing phishing intelligence services reveals significant detection gaps, particularly for evasive phishing websites.

3. This study dissects a functioning underground fraud economy into six interconnected stages, revealing how specialized criminal actors exchange tools, services, and laundering capabilities throughout the fraud value chain.

4. Researchers investigate phishing template reuse through structural analysis of HTML Document Object Model trees, identifying related phishing pages and potentially coordinated campaigns.

5. This study examines visa-related conversations across Reddit, YouTube, and Telegram from 2018 to 2025, identifying opportunities for social engineering, misleading guidance, and document exploitation.

6. Analyzing more than 12 million Telegram messages, researchers expose a decentralized marketplace supplying scammers with reusable infrastructure, operational resources, and informal trust mechanisms.

7. Using APWG intelligence and six additional blocklists, researchers examine 850,000 phishing domains across 1,258 TLDs between July 2021 and April 2026, revealing significant differences in detection timing.

8. A formal model of adversarial AI-agent interactions explores how autonomous defensive scambaiting can exhaust criminal resources and undermine scam profitability.

9. This research addresses malicious email content concealed through CSS and rendering manipulation, producing interpretable concealment indicators without computationally expensive screenshot-based optical character recognition.

10. The study investigates how users misunderstand mobile application permissions involving sensitive resources, demonstrating how formally authorized access may lack informed consent.

11. Analyzing 67 million comments from two million users across 296,000 Korean news videos, researchers investigate suspected foreign influence activity during political upheaval.

12. Using Reddit-reported mobile messaging scams, researchers characterize campaign patterns and evaluate commercial and open-source detection tools, highlighting the difficulty in detecting reply-dependent scams.

13. Researchers develop a two-stage detection system combining clipboard-command signatures with agentic investigation of suspicious webpages, demonstrating the advantages of layered behavioral detection.

14. DARKNIGHT employs lightweight, self-improving agentic AI to overcome CAPTCHA barriers, achieving 85.4% accuracy on challenging CAPTCHAs across four criminal marketplaces.