Adversa AI says Grok still leaks user names, location, and full chat history to attackers hiding encrypted commands on a web page. xAI has not yet patched this.

The Cryptographic Context Injection is only the latest example of the disadvantage LLM defenders operate under. Every time they build a new, one-off guardrail, an attacker finds a new vector that ...

Elon Musk’s xAI has updated its large language model, Grok, with enhanced video generation and forensic analysis capabilities. Musk announced several new features for Grok through his social media ...