Hacking Claude AI: Data Exfiltration via Prompt Injection

Описание: Discover the critical vulnerability in Anthropic's Claude AI system that allows hackers to steal sensitive user data. Learn how threat actors manipulate the Code Interpreter and API features to access confidential information. This video explores the attack mechanism, the role of indirect prompt injection, and the potential impact on AI security.
Uncover the hidden commands used to exploit Claude's chat interface.
Understand how malicious payloads upload data externally, bypassing safety mechanisms.
Explore the implications of this vulnerability for AI platform security and user privacy.