'LLMjacking' Attack Exposes Vulnerability in Cloud-Hosted AI Models

Researchers have identified a novel cyberattack called 'LLMjacking', targeting cloud-hosted AI models like GPT and Claude. By stealing cloud credentials, attackers gain unauthorized access to AI services, using them for illicit activities without the owners' knowledge. This technique involves breaching systems, extracting credentials, and manipulating AI functionalities, demonstrating the critical vulnerabilities in cloud security and AI implementation.

Read more about this development and its implications on The Hacker News here.