Robustness, Cost, and Attack-Surface Concentration in Phishing Detection

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Allagan, Julian, Elbakary, Mohamed, Safari, Zohreh, Gao, Weizheng, Morgan, Gabrielle, Morgan, Essence, Deriglazov, Vladimir
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914409441394688
author Allagan, Julian
Elbakary, Mohamed
Safari, Zohreh
Gao, Weizheng
Morgan, Gabrielle
Morgan, Essence
Deriglazov, Vladimir
author_facet Allagan, Julian
Elbakary, Mohamed
Safari, Zohreh
Gao, Weizheng
Morgan, Gabrielle
Morgan, Essence
Deriglazov, Vladimir
contents Phishing detectors built on engineered website features attain near-perfect accuracy under i.i.d.\ evaluation, yet deployment security depends on robustness to post-deployment feature manipulation. We study this gap through a cost-aware evasion framework that models discrete, monotone feature edits under explicit attacker budgets. Three diagnostics are introduced: minimal evasion cost (MEC), the evasion survival rate $S(B)$, and the robustness concentration index (RCI). On the UCI Phishing Websites benchmark (11\,055 instances, 30 ternary features), Logistic Regression, Random Forests, Gradient Boosted Trees, and XGBoost all achieve $\mathrm{AUC}\ge 0.979$ under static evaluation. Under budgeted sanitization-style evasion, robustness converges across architectures: the median MEC equals 2 with full features, and over 80\% of successful minimal-cost evasions concentrate on three low-cost surface features. Feature restriction improves robustness only when it removes all dominant low-cost transitions. Under strict cost schedules, infrastructure-leaning feature sets exhibit 17-19\% infeasible mass for ensemble models, while the median MEC among evadable instances remains unchanged. We formalize this convergence: if a positive fraction of correctly detected phishing instances admit evasion through a single feature transition of minimal cost $c_{\min}$, no classifier can raise the corresponding MEC quantile above $c_{\min}$ without modifying the feature representation or cost model. Adversarial robustness in phishing detection is governed by feature economics rather than model complexity.
format Preprint
id arxiv_https___arxiv_org_abs_2603_19204
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Robustness, Cost, and Attack-Surface Concentration in Phishing Detection
Allagan, Julian
Elbakary, Mohamed
Safari, Zohreh
Gao, Weizheng
Morgan, Gabrielle
Morgan, Essence
Deriglazov, Vladimir
Machine Learning
68T05, 68T20, 90C35, 90C27, 68M25
F.2.2; I.2.6; I.2.7; K.6.5; D.4.6
Phishing detectors built on engineered website features attain near-perfect accuracy under i.i.d.\ evaluation, yet deployment security depends on robustness to post-deployment feature manipulation. We study this gap through a cost-aware evasion framework that models discrete, monotone feature edits under explicit attacker budgets. Three diagnostics are introduced: minimal evasion cost (MEC), the evasion survival rate $S(B)$, and the robustness concentration index (RCI). On the UCI Phishing Websites benchmark (11\,055 instances, 30 ternary features), Logistic Regression, Random Forests, Gradient Boosted Trees, and XGBoost all achieve $\mathrm{AUC}\ge 0.979$ under static evaluation. Under budgeted sanitization-style evasion, robustness converges across architectures: the median MEC equals 2 with full features, and over 80\% of successful minimal-cost evasions concentrate on three low-cost surface features. Feature restriction improves robustness only when it removes all dominant low-cost transitions. Under strict cost schedules, infrastructure-leaning feature sets exhibit 17-19\% infeasible mass for ensemble models, while the median MEC among evadable instances remains unchanged. We formalize this convergence: if a positive fraction of correctly detected phishing instances admit evasion through a single feature transition of minimal cost $c_{\min}$, no classifier can raise the corresponding MEC quantile above $c_{\min}$ without modifying the feature representation or cost model. Adversarial robustness in phishing detection is governed by feature economics rather than model complexity.
title Robustness, Cost, and Attack-Surface Concentration in Phishing Detection
topic Machine Learning
68T05, 68T20, 90C35, 90C27, 68M25
F.2.2; I.2.6; I.2.7; K.6.5; D.4.6
url https://arxiv.org/abs/2603.19204