Cascading Spy Sheets CSS Fingerprinting
Cascading Spy Sheets: Exploiting Modern CSS for Email and Browser Fingerprinting
topic/papercomputer science/securitycomputer science/webcomputer science/privacy
Paper Summary | NDSS 2025 | Generated by Hermes Agent
Executive Summary: This paper demonstrates CSS-based browser/email fingerprinting that works even when JavaScript is completely disabled — directly threatening Tor Browser and privacy-conscious email clients. By exploiting modern CSS features (custom properties,
@mediaqueries, font metrics, scrollbar styling), attackers can extract system information, track users across sessions, and even leak email content through CSS alone.
1. High-Level Overview
The conventional wisdom: disable JavaScript → stop tracking. This paper shatters that assumption. Modern CSS has become a Turing-complete attack surface capable of exfiltrating data without a single line of JavaScript.
2. CSS Fingerprinting Vectors
- Custom properties &
calc(): Extract computed styles that vary by OS/browser @mediaqueries: Detect screen resolution, color depth, prefers-reduced-motion, etc.- Font metrics: Measure rendered text dimensions to fingerprint installed fonts
- Scrollbar styling: Different platforms render scrollbars differently → OS detection
- CSS animations: Trigger callbacks on animation events (even without JS)
3. Email-Specific Attacks
Email clients often disable JavaScript but render HTML/CSS. CSS-based attacks can:
- Detect which email client is being used (fingerprinting)
- Leak whether an email was opened (read receipts without JS)
- Extract content from the email body via CSS selectors and background-image URLs
4. Impact
- Tor Browser (Safest mode disables JS, but CSS remains active)
- Thunderbird, Apple Mail, Outlook (JS-disabled, CSS-enabled)
- Webmail (Gmail, Outlook.com strip JS but not CSS in many configurations)
5. TL;DR
Modern CSS is powerful enough to fingerprint browsers and leak email content — no JavaScript required. This threatens Tor Browser’s “Safest” mode and all JS-disabled privacy tools.
Sources
- Authors: Leon Trampert, Daniel Weber, Lukas Gerlach, Christian Rossow, Michael Schwarz — CISPA Helmholtz Center