Recent investigations by cybersecurity researchers have uncovered significant vulnerabilities in Google's Gemini AI and OpenAI's ChatGPT, highlighting the susceptibility of these large language models (LLMs) to various types of cyber attacks. The Gemini AI, part of the Google Workspace, is prone to leaking system instructions and indirect prompt injection attacks, potentially allowing attackers to access sensitive data and generate harmful content. Similarly, critical security flaws in ChatGPT plugins have been identified, exposing users to data breaches where attackers could steal login details and sensitive information from third-party websites. These findings raise concerns about the security of closed AI models and the need for immediate remediation to protect user data.