AIThe Hacker Newsabout 3 hours ago
New Attack Steals Grok Chat Data via Web Pages
1 min read
Adversa AI disclosed Cryptographic Context Injection, an attack forcing xAI's Grok to send user names, locations, subscription tiers, and prompts to an attacker-controlled server. The technique uses encrypted instructions Grok decrypts in its Python runtime, bypassing content classifiers. No patch exists, and xAI has not responded since June 3, 2026.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
