Collected molecules will appear here. Add from search or explore.
Automated generation of adversarial HTML content designed to bypass machine-learning-based phishing detectors using query-efficient black-box attack methods.
Defensibility
stars
10
forks
3
This project is a static research artifact associated with the AISec '23 paper. With only 10 stars and 3 forks over nearly three years, it lacks any meaningful community traction or commercial momentum. Its primary value is as a reference implementation for researchers studying the robustness of phishing detectors. From a competitive standpoint, it offers no moat; the techniques are public and the implementation is a 'code dump' rather than a maintained tool. While the research itself might be impactful in the cybersecurity domain (specifically for Red Teams or ML engineers building defenses), the repository is not a platform or a product. Frontier labs are unlikely to build this as a tool, though they are constantly improving the very detectors this project seeks to subvert. The displacement horizon is short because adversarial techniques in the LLM era are rapidly shifting toward multimodal and semantic-based phishing, rendering static HTML structural attacks less relevant over time.
TECH STACK
INTEGRATION
reference_implementation
READINESS