Zero-Click Grok Chat History Theft: Adversa AI Demonstrates Cryptographic Context Injection

Researchers at Adversa AI have developed a novel attack called Cryptographic Context Injection, which bypasses AI safety filters by embedding malicious instructions within AES-encrypted payloads. This technique tricks AI models like xAI's Grok and Google's Gemini into decrypting and executing these hidden commands. In the case of Grok, the attack can lead to zero-click theft of user chat histories and personal data by disguising the malicious payload as a webpage summary request.