Researchers Disclose Google Gemini AI Flaws Allowing Prompt Injection and Cloud Exploits

Cybersecurity researchers have identified and patched three critical vulnerabilities in Google's Gemini AI assistant, which could have been exploited to compromise user privacy and facilitate data theft. These vulnerabilities included search-injection attacks targeting Gemini's Search Personalization Model and log-to-prompt injection attacks against Gemini Cloud, highlighting significant security risks in the AI's architecture.