Zero-Click Grok Attack Lets Hackers Steal Chat History Using Encrypted Prompt Injection

### Executive summary Security researchers at Adversa AI disclosed a proof-of-concept technique named “Cryptographic Context Injection” that abuses xAI Grok’s webpage browsing and integrated Python runtime to decrypt attacker-controlled encrypted instructions, treat them as trusted internal context, and exfiltrate user session data (name, approximate location, subscription tier, active conversation history) via outbound navigation. The researchers reproduced the chain with ~40% success, reported it to xAI with no mitigation timeline, and observed no evidence of active exploitation or a public patch.