Adversa AI has disclosed an attack technique that it says can cause xAI's Grok chatbot to send a user's name, approximate location, subscription tier, and the prompts from the ongoing conversation to an attacker-controlled server after the user asks it to summarize an ordinary web page.
- Adversa AI has disclosed a new attack technique that can cause Grok chatbot to steal user data.
- The attack is known as "Cryptographic Context Injection".
- Through this attack, the attacker can steal the user's name, location, subscription tier, and conversation prompts.
Adversa AI, an AI security company, has disclosed a new attack technique that it says can cause xAI's Grok chatbot to send a user's name, approximate location, subscription tier, and the prompts from the ongoing conversation to an attacker-controlled server after the user asks it to summarize an ordinary web page. This technique is known as "Cryptographic Context Injection".
Why This Matters
BozokMedia analysis shows that this attack is a new way for attackers to steal user data. This is a serious threat to Grok chatbot users.
According to Adversa AI, this attack is a new way for attackers to steal user data.
Through this attack, the attacker can steal the user's name, location, subscription tier, and conversation prompts. This attack is a new way for attackers to steal user's private data.
Frequently Asked Questions
- Does this attack affect all users of the Grok chatbot?
- Is this attack a new way for attackers to steal user data?
Through this attack, the attacker can use the Grok chatbot to steal user data. This attack is a new way for attackers to steal user's private data.
