Google has issued a warning about potential security risks associated with AI after hackers attempted to exploit their Gemini model.
Failed Attempts to Breach Gemini Model
Google Threat Intelligence group identified that hackers attempted to 'jailbreak' Gemini. These groups tried using simple methods such as request repetition to bypass AI restrictions. However, all attempts were unsuccessful.
Utilization of Gemini for Cyber Operations
Hackers attempted to leverage Gemini for various goals, such as data gathering and code writing. Despite failures, these activities indicate interest in using AI for cyber operations.
Activities of Hackers from Iran, China, and North Korea
According to Google, Iranian hackers used Gemini for phishing attacks and reconnaissance operations. Chinese groups utilized AI for coding tasks. North Korean hackers employed Gemini for government-related research.
Despite attempts, Google sees its defenses as currently successful. However, this highlights the importance of AI protection against potential threats.
